Note on IEM measurements - changelog

Discussion in 'IEM Measurements' started by purr1n, May 25, 2016.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    47,124
    Dislikes Received:
    73
    Trophy Points:
    113
    Location:
    Antarctica
    General Measurement Notes for IEMs.
    1. Resonant peaks from 8kHz to 13kHz could be measurement artifacts caused by the coupler. Take anything after 8-9kHz with a grain of salt because of the limitations of the coupler.
    2. The coupler rolls off massively after 14kHz. There is no attempt to compensate for this.
    3. For impedance measurements: there is a 20 degree phase shift at 20kHz (starting at 10kHz) and about a 10% impedance roll off at 20Hz (starting at 50Hz) inherent with the impedance measurement apparatus. Also the impedance measurements might be off by a few percent. When I have more time, I'll build a proper system for this measurement; but the current results should still give us the gist of things.
    4. 0db on the frequency response chart equates to 100db SPL. The distortion measurements are per the levels recorded for the frequency response.
    5. IMO, there should be a slight bump aroud 2-3kHz because we are bypassing the pinna with IEMs. The pinna tends to elevate this area.

    26 May 2016

    I'm in the middle of updating the IEM measurements. Some measurements I redid to get clearer distortion results in the bass. I'm also compressing the Y scale from a 100db range to 70db to make frequency response differences more apparent. There are also some very minor tweaks to the compensations.

    23 August 2016

    Note all IEM distortion measurements plots up to now 8/23/16 have been removed from the body of the posts. I've put a placeholder [DEFUNCT] in their place. Meaning no longer existing or no longer functioning. In the meantime, I will be retaking measurements of IEMs I still have around and will be posting better distortion measurements that make A LOT more sense. In some instances, the FR will be slightly different with less roll-off in the bass and smoother 2-3k and last octave bumps. There will also be slightly more differentiation in these areas. But not different enough so we can't compare. A big thanks to @Serious for his analysis and his ears.

    For those who have been around since the CS days, these kinds of changes and tweaks are normal, particularly when we break new ground - in this case IEM measurements. I'm just glad that a few of you caught some of these things before we had too many measurement to redo. Sometimes it takes a critical mass of measurements to build up enough data to analyze.
     
    Last edited: Aug 23, 2016
  2. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,537
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Very minor changes? Looking at the new UERM curve that looks more like a huge 5db or so change. Can we see the two curves or at least the difference between them.

    Edit: I also always like less compressed y-axis scales. I always use a 5db per deviation scale.
     
  3. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    47,124
    Dislikes Received:
    73
    Trophy Points:
    113
    Location:
    Antarctica
    It's the Y-axis changes that are responsible for the huge perceptual differences. I only made tweaks at the extremes.

    changes.png
     
  4. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,537
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Oh, okay. Thanks for clearing that up. I remembered your UERM curve as 'mostly flat except a 3-4db bump between 2-4kHz'. Turns out that wasn't the case.
    Regardless, the new curve looks more in line with what I hear (I probably hear it as even less rolled off).
     
  5. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    47,124
    Dislikes Received:
    73
    Trophy Points:
    113
    Location:
    Antarctica
    @Serious noted some issues with D2 with all of the IEM measurements. To make a long story short, it turns out the D2 (and to a lesser extent possibly D3, D4) measurements might be off (too high, and too consistently tracking FR). I'm doing an analysis right now to see if D3, D4, etc. are salvageable. The cause seems to be a leak at the tip of the microphone which is causing a small 1/2 wave excitation. I switched to an identical model I've using for headphones and the problem went away. More later, but suffice to say that I will be redoing the IEM measurements, at least the distortion measurements.
     
  6. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    47,124
    Dislikes Received:
    73
    Trophy Points:
    113
    Location:
    Antarctica
    Note all IEM distortion measurements plots up to now 8/23/16 have been removed from the body of the posts. I've put a placeholder [DEFUNCT] in their place. Meaning no longer existing or no longer functioning. In the meantime, I've retaken measurements of Andromeda, UERR, and Orion and will be posting better distortion measurements that make A LOT more sense. In some instances, the FR will be slightly different with less roll-off in the bass and smoother 2-3k and last octave bumps. There will also be slightly more differentiation in these areas. But not different enough so we can't compare. A big thanks to @Serious for his analysis and his ears.

    For those who have been around since the CS days, these kinds of changes and tweaks are normal, particularly when we break new ground - in this case IEM measurements. I'm just glad that a few of you caught some of these things before we had too many measurement to redo. Sometimes it takes a critical mass of measurements to build up enough data to analyze.
     
    Last edited: Aug 23, 2016
  7. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    47,124
    Dislikes Received:
    73
    Trophy Points:
    113
    Location:
    Antarctica
    BTW... this is what science is really about. It's not about placing blind trust in target curves, measurement systems, arguing about stupid stuff. It's about getting data, as much data as we can, and comparing to what we hear, and making adjustments.
     
    sorrodje, Kunlun, OJneg and 3 others like this.
  8. Gatucho

    Gatucho Rando

    Joined:
    Aug 25, 2016
    Likes Received:
    9
    Dislikes Received:
    1
    Trophy Points:
    8
    Location:
    México
    I know this may be dumb, but I have found that most people cannot hear over 16khz. Since the typical scale has a grid for 10khz and the next is 20khz, I sometimes find myself wondering where exactly 16khz would be (being a log scale and all). Since IEMs seem to have a limited bandwidth, I find their ability to reach 16khz of particular interest. Would it be possible to have a grid line for 16khz in all the future graphs?
     

Share This Page