ETA Mini Analysis and Measurements

Discussion in 'Headphone Measurements' started by Vtory, Feb 19, 2022.

  1. Vtory

    Vtory Audiophile™

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    9,939
    Trophy Points:
    113
    Location:
    East Coast
    Relevant Links

    I finally get to take measurements of ETA Mini headphones. Do these subjectively good-sounding headphones measure well, too? Let's find out.

    [​IMG]

    Above is FR plot compensated against older SBAF curve which I consistently use for all my measurements. 1/24 oct smoothed. My favorite pads (Geekria L) were equipped.
    • Mild imbalances were seen in the graph, but nothing looks critical to me.
    • The overall tonal balance seems shaped like V centering at 900hz. That's largely consistent with what I heard.
    • Bass starts to roll off at 60hz, but please note that in reality this hugely depends on how earpads are fitted with ears (real/dummy). This implies while the result above wasn't too different from my perception, you may hear a little deeper or shallower basses based on how the Mini seals your ears.
    • Between 4khz and 5.5khz, energy level looks higher than the surrounding regions. Also a little zigzagging to my eyes. Subjectively my ears didn't find it rough at all, but some pads helped to smooth out it measurably. More on this below.
    • Big scoop around 7-10k seems more drastic than what I perceived.
    [​IMG]

    Pre-post comparative plots above. Solid line means pre-EQ response and dotted one corresponds to post-EQ.
    • My filters largely "weaken" V-ness by dropping some bits in whole bass and upper treble.
    • Post EQ, midrange level is more in line with bass and treble energies.
    [​IMG]

    Differences across three pads I tested. Best orientation for each was displayed (backward for Geekria L and forward for the rest). For visual convenience, I zoomed in a little bit in y-axis.
    • Gerod L (green) does smooth out treble response between 3khz and 6khz. I doubt if I can hear it as much as the graph suggests though.
    • Geekria L (red) seems the most neutral among the bunch.
    • Measured differences were narrower than I heard/thought. I surmise that maybe different physical texture or wearing experience served as some confounders in my perception.
    More to come in the next post..
     
    • Like Like x 4
    • Epic Epic x 1
    • List
    Last edited: Feb 19, 2022
  2. Vtory

    Vtory Audiophile™

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    9,939
    Trophy Points:
    113
    Location:
    East Coast
    Now, let's talk about non-linear measurements. Needless to say, they need way more caution than linear results. False positives and false negatives are more likely imho. Measurement correctness aside, audibility is controversial at best, too.

    [​IMG]
    [​IMG]
    [​IMG]

    From top to bottom, D2, D3, and D4 plots for three different sweep signal levels (calibrated at 300hz). Left channel is only displayed (as Ev said the specific demo pair sounded a little better in left). But I'd comment right ch results were more or less the same in my measurements.
    • Mini's V-shape response affects fundamental levels. Be aware of varying "denominator".
    • 2khz pattern in D2 seems tricky. Maybe that's somewhat related to resonance in earcanel and/or eardrum.. but I don't have any concrete explanation atm.
    • Louder signal looks like meaningful burden for Mini's drivers in terms of driver excursion. But rather understandably so.

    [​IMG]
    [​IMG]

    Finally CSD results for both channel with the signal calibrated at [email protected]
    • Results look a little cleaner than Gen-G 1.1 or ESX 900 I previously measured. But because mini is structurally more open than those two, interpretation needs more caution. Better not over-praise nor overinterpret.
    • Midrange contains more energy than Grado or Focal. Not surprising for airflow control and acoustic impedance of Mini (kinda common tuning shown in all ETA hps).
    To conclude, Mini shows decent performance objectively to say the least. In every category, I couldn't find anything critically wrong. Also happy to see the results support what I found in subjective evaluation to some extent (60% or so?). Good job, ETA team!
     
    • Like Like x 1
    • Epic Epic x 1
    • List
    Last edited: Feb 20, 2022
  3. spoony

    spoony Spooky

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    606
    Trophy Points:
    93
    Cool! May I ask to see the FR graphs overlaid with some of the usual suspects? (Grado Hemp, Senns)
     
  4. Vtory

    Vtory Audiophile™

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    9,939
    Trophy Points:
    113
    Location:
    East Coast
    Sure things. Here you go! All without any DSPs btw.

    [​IMG]

    [​IMG]
     
    • Like Like x 3
    • Epic Epic x 3
    • List
  5. spoony

    spoony Spooky

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    606
    Trophy Points:
    93
    Thank you!
     
  6. Vtory

    Vtory Audiophile™

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    9,939
    Trophy Points:
    113
    Location:
    East Coast
    Socks-in Measurements.

    Q: What does it mean?
    A: Refer to this post: https://www.superbestaudiofriends.o...essions-and-reviews-thread.11855/#post-375169

    [​IMG]

    Above is the comparison across different "outer" pads. All normalized at 1khz.
    • Geekria forward vs backward is mostly coming from the treble differences. Forward direction functions like some kind of high-shelving filter.
    • Gerod forward makes the balance more V-ish. I haven't measured the backward direction, but as far as I heard, similar treble reduction seems to occur.

    [​IMG]

    Fixing Geekria-L as outer pads, the graph above presents the incremental effect of the socks-in.
    • Socks-in generally makes responses quite less smooth between 1khz and 4khz.
    • There must be a constructive interference at 2khz and destructive one at 2.4khz. As the measurement can only capture overall sum, I can't figure out underlying wave components for the interference process.
    As another data points, adding CSD plots for the Socked Geekria-L Bwd below for the food for thoughts.

    [​IMG]
    [​IMG]
     

Share This Page