Harmonic Distortion Measurements / Input Gain

Discussion in 'Measurement Setups, Systems, and Standards' started by Klasse, Feb 25, 2016.

  1. Klasse

    Klasse Friend

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    117
    Dislikes Received:
    0
    Trophy Points:
    43
    I've been measuring some headphones using an extremely cheap-skate setup
    (CompactDisc, Laptop's integrated mic, ARTA (demo) and a bloody exponential compensation to match Marv's measurements)

    Surprisingly I get very consistent FR measurements as exposed here: Late T1 v1 (+ HD650) Measurements

    Now, harmonic distortion measurements are a whole different story.

    When we normally expect something like this (Marv's plot):
    [​IMG]

    I'm getting something like this:
    [​IMG]

    Or this:
    [​IMG]

    As you can see, nice FR but crazy, spiky and inconsistent HD.
    First I thought, ok... cheap Mic, not surprised.

    Then I saw @philipmorgan 's measurements here:
    Request: Study of Damping Factor and Planar Magnetic Headphones

    He build up a more serious measurement rig using a real (Dayton) mic.

    His HD650 measurement:
    [​IMG]

    Better than mine, but still spiky 2nd Harmonic Distortion as you can see...
    (Very different from Marv's and others)

    So, what's up with our measurements? Are we missing something?
    What considerations one should have in mind when trying to measure Harmonic Distortion?

    Thanks in advance,
    Klasse
     
    Last edited: Feb 25, 2016
  2. OJneg

    OJneg The Most Insufferable

    Friend
    Joined:
    Sep 25, 2015
    Likes Received:
    3,253
    Dislikes Received:
    13
    Trophy Points:
    113
    Location:
    Chicago
    Looks to me like you're not managing levels well. Noise floor is usurping the THD measurement. How loud is the headphone running when you do sweeps?
     
    Klasse likes this.
  3. Klasse

    Klasse Friend

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    117
    Dislikes Received:
    0
    Trophy Points:
    43
    You are the man!

    Louder:
    [​IMG]
    Still not Messiah-quality, but much more coherent.

    My healthy customs betrayed me this time.
    I normally listen to music at pretty low levels so I've used to set the levels accordingly.

    I guess @philipmorgan should be able to match Marv's Harmonic Distortion measurements raising the volume.
     
  4. philipmorgan

    philipmorgan Member of the month

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    2,073
    Dislikes Received:
    1
    Trophy Points:
    93
    Location:
    Sebastopol, CA
    Home Page:
    When doing the measurements @Klasse cited, I was aiming for a 90db peak area on FR curve. Is that not loud enough?

    I haven't been able to find any best practices (static.changstar.com's been spotty the last week or two) about this.

    I'm rebuilding my measurement rig this weekend so that they stress the headphone bands less. I'll try measuring at a louder volume then to see how distortion changes and if I can replicate others' results more closely.
     
    Klasse likes this.
  5. OJneg

    OJneg The Most Insufferable

    Friend
    Joined:
    Sep 25, 2015
    Likes Received:
    3,253
    Dislikes Received:
    13
    Trophy Points:
    113
    Location:
    Chicago
    You need to not only measure at a loud enough level to excite the headphone's distortion, but you also need a recording chain (mic-pre-A/D) that has low enough noise and distortion so you have enough dynamic range. Think about it has having enough "room" underneath the signal.

    So you could very well have been playing at 90dB SPL, but your gain settings were not set right.
     
    philipmorgan and Klasse like this.
  6. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend
    Joined:
    Sep 24, 2015
    Likes Received:
    37,942
    Dislikes Received:
    59
    Trophy Points:
    113
    Location:
    Calabasas, CA
    In terms of loudness registered in ARTA (or whatever app) you are using for measurements, it can vary depending upon measurement method, calibration of mic, weighting, etc. For my plots, the 0db point equates to 100db. I generally try to align the 500-1kHz area to 90-87db.

    There are two places to adjust input gain. It's matter of playing with it to get the lowest noise results:
    1. Microphone preamp
    2. Windows Control Panel - Sound - Recording Devices
    Also be sure to check sampling frequency. MLS, sweeps, noise OUT should be the same as the signal IN.

    For obvious reasons, harmonic distortion measurements should always indicate reference SPL. I know 90db is very high for normal listening, but I like to artificially bump it up this high because it makes differences easier to see, and also because even at this higher level, there is correlation to what is actually heard at normal listening levels.
     
    Last edited: Feb 26, 2016
    Klasse and philipmorgan like this.

Share This Page