Headphone Amplifier Distortion vs Amplitude Technical Measurements

Discussion in 'Headphone Amplifier Measurements' started by atomicbob, May 29, 2022.

  1. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Sep 27, 2015
    Likes Received:
    Trophy Points:
    On planet
    Amplifier Distortion vs Amplitude Technical Measurements

    In this first installment three low cost schiit headphone amplifiers are compared.

    Test conditions:
    Selectable gain set to high
    Amplifier gain adjusted for 0 dB input to output
    Load 300 Ω
    Ultron 7DJ8 tube in Vali 2+

    01 magni heresy 3+ amps cropped.jpg
    Magni Heresy top
    Magni 3+ bottom

    More measurements:

    Magni 3+

    02 vali_2+_cropped_small.jpg
    Magni Vali 2+ with Ultron 7DJ8 tube

    More Vali 2+ measurements:

    03 Negative Feedback.png
    Amplifier Negative Feedback

    Negative FeedBack (NFB) involves subtracting a fraction of the open loop Amplifier gain stage output from its input.
    That fraction is shown as β in the diagram above.

    β = NFB dB and the terms will be used interchangeably.

    NFB dB values reported in the following graphs are approximate as they are a simplification from gain-bandwidth plots.

    04 202200514 vali 2+ 1 KHz distortion vs amp A=0 dB high gain 300R annotated trace ID.png
    These distortion sweeps were performed over the range of +20 to -20 dBu.
    FFT trace is distortion spectrum at 0 dBu.
    Bandwidth for distortion observations is 22 Hz to 22 KHz unless otherwise noted.

    Trace ID:
    Yellow: THD+N
    Green: D2
    Tan: D3
    Red: 4+HD+N with bandwidth 400 Hz to 22 KHz
    Blue: FFT distortion spectrum at 0 dBu

    4+HD+N is also called Crap Factor™ and the terms are used interchangeably.

    A tutorial on these graphs available here:

    05 202200514 vali 2+ 1 KHz distortion vs amp A=0 dB high gain 300R annotated.png
    Vali 2+ distortion vs amplitude

    Many headphones have sensitivity such that 0 dBu will produce approximately 100 to 105 dB SPL. While interesting to observe amplifier behavior above 0 dBu the range of significant interest is below 0 dBu.

    Note the yellow box in the graph above surrounding Red Crap Factor™ trace -20 to 0 dBu.
    Also note NFB dB = 8 for the Vali 2+.

    06 202200514 magni 3+ 1 KHz distortion vs amp A=0 dB high gain 300R annotated.png
    Magni 3+ distortion vs amplitude

    While the THD+N is very different from that of the Vali 2+ observe the CF is nearly identical. Yet the Magni 3+ uses 30 dB of NFB.

    07 202200514 heresy 1 KHz distortion vs amp A=0 dB high gain 300R annotated.png
    Magni Heresy distortion vs amplitude

    THD+N and CF are nearly the same as Magni 3+, yet Heresy has 115 dB of NFB.

    08 vali 2+  magni 3+  heresy  distortion animation.gif
    Animation of the previous three measurements

    Again, observe how CF is nearly the same for all three but considerably different amounts of NFB are utilized for each amplifier design.

    Several key points to remember:
    1. THD+N masks differentiating distortion components
    2. Each amp has different D2 / D3 sweep signatures
    3. THD+N @ 1 KHz is 0.00216% for Heresy and 0.00252% for Magni – very similar – yet sweeps demonstrate clear differences between them
    4. 4+HD+N is nearly identical for all three amps in the range of interest

    SBAF preference poll of these three headphone amps may be found here:

    Between these measurements and the SBAF poll one might infer a preference exists for amplifiers with lower NFB needed to achieve a particular CF distortion performance.

    20220529 Poll Results to date:
    09 202200529 poll results.png

    20220530 edited to correct THD+N typos in key point values.
    • Epic Epic x 20
    • Like Like x 7
    • List
    Last edited: May 30, 2022
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Sep 24, 2015
    Likes Received:
    Trophy Points:
    Padre Island CC TX
    Very nice and scientific! I'm actually somewhat surprised that Vali 2+ was the first choice, I thought it would be more even, closer to 50/50 split.

    Magni Heresy has two stages and doesn't employ global feedback. However each stage uses an opamp. The output stage is a bunch of paralleled opamps in unity gain which means 115db of feedback. Same for the first gain stage, when set for unity gain (best measurements), 115db of local feedback there too!

    Magni 3+ has local feedback in the output stage I believe. Probably something like 22db I am guessing.

    Pretty sure people buy the Heresy because of the black.
    • Like Like x 1
    • Agreed, ditto, +1 Agreed, ditto, +1 x 1
    • List
  3. Royaume

    Royaume Acquaintance

    Apr 6, 2019
    Likes Received:
    Trophy Points:
    Kansas, US
    Awesome work! I simply love reading your technical insights!
    Since discussing Cheever's paper on NFB, I have taken on board what you taught me, and I have decided to leave the investigation of amplifier architecture and design to the experts such as yourself. I learn a lot from reading your thoughts and analysing your measurements.
    Please keep it up :bow:

Share This Page