1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Did I just measure the dreaded Sabre "Glare"?

Discussion in 'Sound Science' started by pwjazz, Apr 28, 2018.
  1. pwjazz
    Looking for a DAP with a nice built in amp, I recently acquired a new overstock LG V20 on eBay. The few negative reviews I've read of it talk about the ESS Sabre "glare", describing it as either a bright (specifically sharp) frequency response or a sort of digital or unnatural sound. The only other Sabre device I ever had was a Dragonfly Black, and it's too long ago for me to remember anything reliably about how it sounded, though I want to say that I didn't love the highs on that either.

    Well, I plugged my already bright DT 1990s into my V20 and listened to a bunch of stuff. Especially on things like the 4th movement of Beethoven's 9th which starts out with some high and loud horns and strings, high frequencies on the V20 did indeed sound remarkably unpleasant. The strings sounded scratchy and the horns almost as if they came out of a synthesizer. By comparison, things sounded much better out of my Topping NX4, with a more pleasant and natural tone.

    Eager to see whether this was expectation bias, I decided to take some measurements with my MiniDSP E.A.R.S. These measurements follow the manual, i.e. use Room EQ Wizard, volume match a 300Hz sine wave at 84dB(C) S, etc. In the past, when I've had developed subjective listening impressions about differences in amps and DACs, I've never been able to find any big difference in my measurements. This time was different!

    I tested three different scenarios:

    Macbook USB -> NX4 -> DT 1990
    LG V20 -> DT 1990
    LG V20 -> Line in on NX4 -> DT 1990

    I ran the measurements twice for each scenario to see if there were any major variations between runs. W.R.T. frequency response and impulse response, the results were consistent across the two runs.

    Full images are available on Imgur.

    First let's talk about what's not different - frequency response.


    Yeah there's a slight difference above 17.5 kHz, but I can't hear that high anyway (14 kHz is my limit).

    What is different (pretty radically so) is the impulse response.


    Two things strike me about this:

    1. The LG V20 has much faster initial decay after reaching 100% of signal

    2. The initial response is actually a -13% (out of phase?) before it reaches 100%. I haven't taken a whole lot of such measurements, but I think this is the first time I've seen this kind of pronounced dip. Incidentally I also measured my LCD2C out of the LG V20 and observed a similar dip. Furthermore, if I amp via the NX4, I still see the same dip, so I'm pretty sure it's the V20's DAC and not the amp that's responsible.

    The faster initial decay seems like it probably just improves clarity. What's interesting to me is that initial 13% dip.

    My question to those of you who know more about reading impulse responses... Could that weird initial dip in the V20's response be what's to blame for higher frequencies sounding unnatural and if so, might we expect to see similar dips when measuring headphones with other Sabre devices? Is this what the Sabre glare looks like when measured?
  2. castleofargh Contributor
    you need to be cautious with impulse interpretation as it contains a lot of information and is a lot more than just the intuitive little "how fast does it move? how fast does it stop?"
    different DACs will have different filters, oversampling, whatever... that will result in different impulses. different FR, damping ratios from different amps, reverb, etc. they all end up showing up on the impulse.

    here are 2 sources directly in a loop DAC/amp to ADC, I guess I can play with the E.A.R.S and a headphone if needed, but I believe this already conveys my point well enough. you see all that crap, yet I honestly can't tell them apart in blind test with music I'm used to at reasonable listening levels if I send the Odac into the UHA760's amp. I guess I'm time smearing insensitive :deadhorse:

    your impulses are honestly very similar and I wouldn't try to look too much for subjective interpretation into them.
    RRod likes this.
  3. SilverEars
    Wow, this is something new to me. Different impulse response from different sources.
  4. Glmoneydawg
    More likely amp section related than digital
  5. SilverEars
    Tyll's measurements includes this, and talks about the headphone's impulse response. What is really the ideal reference if there is such differences?
  6. Glmoneydawg
    No frequency response difference rules out digital issue.....rìght?Just asking,not my strong point here:)
    Last edited: Apr 28, 2018
  7. SilverEars
    It would be interesting to see impulse repsonses with controlled measurments.

    1. Same headphone and DAC, different amps

    2. Same headphone and amp, different DACs

    FR just shows tonal response
  8. pwjazz
    That would be awesome! I've always understood the headphone's characteristics to dominate, so even if two DACs measure very differently unloaded, it wouldn't surprise me that the same headphones measure basically the same from both.
  9. pwjazz
    As mentioned, I tested:

    Macbook USB -> NX4 -> DT 1990
    LG V20 -> DT 1990
    LG V20 -> Line in on NX4 -> DT 1990

    This hits both of your suggested conditions. Basically, the amp had a bit of an effect on the decay post peak, but the leading dip (or no dip) seemed completely DAC dependent.

  10. pwjazz
    I controlled for this, and that initial dip doesn't change with a different amp.
  11. SilverEars
    More measurements you do under various controlled conditions, it's more reliable and more confidence level in patterns. Redoing them as well to see if getting consistant measurements.

    Castleofargh's graph show significant differences. If that is out of the same headphone, how can one reliably put out impulse response and claim that impulse response is solely a characteristic of the headphone?
    Last edited: Apr 29, 2018
  12. castleofargh Contributor
    my graph didn't involve a headphone and mic, I posted it only as an extreme example of how some of the changes in an impulse can look drastic visually and yet not make that much of a subjective change. of course other changes in an impulse totally do! just not what's on those 2 specific impulses. at least not for me.
    with a headphone and mic involved, as pwjazz suggests, the headphone's own movements become the main event by a significant margin. remember my graphs look like big scary changes but are really just ringing from the filter at around 20khz being caused differently(also this was done at 44.1khz for even scarier ringing like in marketing BS "demonstrations" with IR).

    look I make no promises because I'm lazy so not making them saves me from breaking them all the time^_^, but next week they've announce some rainy days. if I get stuck at home and remember, I'll run a bunch of combinations to show practical stuff similar to what @pwjazz did just for good measure.
  13. pwjazz
    That would be amazing, no pressure though. I'm just hear to learn and I'll take whatever I can get. I'm hopeful that over time we'll see more measurements of headphone/amp/dac combinations, as traditionally I've seen either measurements comparing just DACs or amps, or measurements comparing just headphones (sometimes with no reference to the specific DAC or amp used). Some people swear by their DACs and/or amps noticeably impacting sound, some say that all well implemented sources sound the same unless they're specifically trying to alter the sound. I suspect that the truth is somewhere in-between, and I'm hopeful that measuring will tell us something.
  14. pwjazz
    Gotcha. I haven't done a ton of these yet, but I've done a few.



    The difference in these cases seem tiny compared with what I saw in my most recent test. One difference could be that testing the LG V20 required me to play a pre-recorded frequency sweep through a media player on the phone (USB Audio Player Pro). I'll try feeding the NX4 via USB from the LG V20 and UAPP and see how that looks.
  15. SilverEars
    And I thought it was for headphones out of different sources... :expressionless:

Share This Page