Weird issue with my waterfall plots. USB DAC or MIC related?

Discussion in 'Measurement Techniques Discussion' started by AverageJoeAudiopihle, Jul 31, 2017.

  1. AverageJoeAudiopihle

    AverageJoeAudiopihle New

    Joined:
    Jul 11, 2017
    Likes Received:
    17
    Trophy Points:
    3
    Location:
    California
    Home Page:
    http://imgur.com/a/4cGbS

    Gallery of 3 recent measurements. Done on tweeter axis, outdoor at 1m, tweeter is 82 inches in the air. MiniDSP usb Mic, Behringer U-Control UCA202 as USB DAC

    I don't normally do waterfall plots when I do a review, but I wanted to start incorporating them in the future. Except every measurement I've taken shows almost perfectly matching slow decay from about 6khz to 20khz.

    Because the problem is so uniform, I'm assuming it some hardware issue. Has anyone seen anything like this before?
     
  2. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    You have to change the CSD settings in REW. Window rise-time, window length, what kind of IR window you use, both left and right of the IR peak. The stock settings I believe are made to really show room modes in the bass region, but they don't work for the smaller time-scale of anechoic measurements or headphone measurements.

    I think I described most of my settings here: http://www.superbestaudiofriends.or...nt-setups-post-your-rig.141/page-7#post-50142
     
  3. BobApelsin

    BobApelsin New

    Contributor
    Joined:
    Jul 25, 2017
    Likes Received:
    1
    Trophy Points:
    3
    Location:
    Sweden
    Hi about uca 202 /222 ( only a external colour diff)

    Technical and modification to circuit needed so if not DIY happy avoid it.

    It seems those units have a hardware problem with leaking overhearing/crossfeed back from output to inputsignal.
    If strong signals and / or using external amp then better readings

    (if you can read swedish they have an extensive explanation on diy forums.
    http://www.faktiskt.se/phpBB3/viewtopic.php?f=3&t=65974 )

    But mostly overhearing troubles in bass area measuremants from 20hz and slopes down to 5k .
    -40db as worst and fixed it goes down to -88 ish db.
    Floating/virtuell ground issue with the headphone amp output that creates small feedback to vitrual earth.

    Your tweeter range measuremnts /waterfalls are above that , after 5k its around 6db improvement after modding the circuit.

    Dont know how much the mod will affect your measurements .

    Some using Focusrite 2i2 instead.

    Regards magnus
     
  4. AverageJoeAudiopihle

    AverageJoeAudiopihle New

    Joined:
    Jul 11, 2017
    Likes Received:
    17
    Trophy Points:
    3
    Location:
    California
    Home Page:
    Yeah, for this example it's the turnkey 125ms window lengths on both sides
     
  5. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,976
    Trophy Points:
    113
    Location:
    Vancouver BC
    What length measurement sweep are you using? Note in @Serious 's measurement settings he mentions he uses a short sweep for CSDs because a long one changes the phase and screws up the CSD.

    This happens with one of my mic interfaces too: the long (1M) sweep recommended for distortion measurements screws up the impulse response, giving a bizarre, high-amplitude pre-ring, a phase reversal at 0 sec, and a waterfall that doesn't match the FR curve at zero time (and reversing the phase, which can be done in the REW measurement preferences, doesn't cure this). A 128k sweep, though, gives a 'proper' IR and a consistent zero-time waterfall. So changing the sweep time, as well as the other settings, might help.

    One more thing about the UCA202: if you're going to use it for headphone measurements, don't use its headphone-out as the source - the impedance is about 50 Ohms.
     
  6. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    134
    Trophy Points:
    33
    Are your output and input on the same soundcard? If they are different devices and are operating at different sample rates (even if nominally the same) a longer sweep will expose the sample rate differences, particularly if they are configured in the OS to run at diffreent rates and the OS is resampling the data for one of them.
     
  7. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,976
    Trophy Points:
    113
    Location:
    Vancouver BC
    Output is analogue (from a cheap portable USB DAC/amp). The input device that changes impulse behavior with sweep time is the ATR2USB. Interestingly, an even cheaper unit doesn't do this.

    Sample rate is set to 44.1 in the REW soundcard prefs (I presume that sets both out and in?) and in the Windows mic settings.
     
  8. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    134
    Trophy Points:
    33
    REW's sample rate is used for input and output, but it cannot control the rate at which devices actually operate. It asks for lines at the required rate and usually the OS will provide them, but the OS may be resampling data to/from the devices if they are configured in the OS to operate at a different rate.
     
  9. Biodegraded

    Biodegraded Friend

    Pyrate Contributor
    Joined:
    May 28, 2017
    Likes Received:
    7,976
    Trophy Points:
    113
    Location:
    Vancouver BC
    As an experiment, I changed everything (audio output, REW, mic input) to 48k, the highest rate supported by both my audio DAC and the mic interface (and what Windows would resample to if it was doing it on the sly). Same result.

    Could it be latency in the mic interface? I'm not using a timing reference.
     
  10. JohnM

    JohnM Author of REW - Rando

    Joined:
    Dec 13, 2015
    Likes Received:
    134
    Trophy Points:
    33
    Latency wouldn't usually be an issue with that kind of setup. You can check by looking on the Scope graph after you make a measurement, that graph has the time signal for the last measurement made so you could see on there if the capture were truncated in some way. If you make a measurement with 256k and 1M and attach (or email me) the .mdat file I can take a look and see if anything leaps out.
     

Share This Page