Merv's Politically Incorrect Audio Blog

Discussion in 'SBAF Blogs' started by purr1n, Dec 26, 2018.

  1. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    8,904
    Dislikes Received:
    3
    Trophy Points:
    113
    Location:
    On planet
    Two reminders:
    Tutorial - Interpreting THD and THD+N specifications and graphs

    Tutorial - A tale of 6 amps THD+N and THD specifications

    Chasing numbers is a fool's errand. They are lossy compression equivalent for technical measurements with an enormous quantitative data reduction.

    Consider the following three amps THD+N, taken from their respective Left channel technical measurements posts:

    THD+N Left channel
    0.002% Liquid Platinum
    0.006% Lyr3
    0.009% DSHA-3F

    Those numbers are all respectably low and don't appear to differ by much.

    Now consider 50+7000Hz FFTs for the same three amps at approximately +3 dBu:

    50+7000 Hz Liquid Platinum
    04 20181110 LiqPlat 50+7000Hz 0dBu 300R 4XF.png

    50+7000 Hz Lyr3
    05 20180430 lyr3 50+7000Hz 0dBu 300R 4XF.png

    50+7000 Hz DSHA-3F
    06 20190221 DSHA-3F 50+7000Hz 0dBu 300R 4XF.png

    Those three amps have quite different FFT signatures.

    Measurements matter, when you know which ones are worth comparing. When chasing numbers the assumption is lower is better. I don't believe that is a good protocol. Looking at FFT signatures one might wish to find amps with signatures aligning with personal preferences. A superior protocol.
     
    Lyer25, Syzygy, geniekid and 14 others like this.
  2. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    45,047
    Dislikes Received:
    70
    Trophy Points:
    113
    Location:
    Antarctica
    CLASSIC SOUND DACS, MODERN SOUND DACS, AND DACS IN-BETWEEN

    This is a cut-and-paste from another thread with additional commentary. I felt this belongs here because a few people said it should be sticky and I've felt a need to refer to it in my DAC impressions. The answer goes beyond the original question below.

    Question: Do you think Schiit Multibit DACs are the only good DACs that exist?
    Answer:

    I'll answer this. Some history is required. This is the very long version which details my search for DACs many years ago: https://www.head-fi.org/threads/tho...-just-to-get-you-to-think-about-stuff.693798/

    Note that some people call me a Schiit shill, but this thread details my DAC journey starting in 2013, way before Yggdrasil even existed! The fact is, I stumbled (or re-stumbled) into the Theta DAC brands where Mike Moffat happened to be the designer. The entire Schiit mega-combo-burrito filter thing with its holographic imaging was already invented in the 1990s on the Theta Gen V. It should be noted that knew nothing about the Schiit filter until I sought out Mike Moffat @baldr - it was not the other way around - keep in mind that Changstar.com was avoided by manufacturers like the plague. I begged Mike to bring something like the Theta Gen V back. As an aside, if you use headphones, you are only getting 60% of what the Schiit Multibit DACs can do. Only speakers with good dispersion and off-axis characteristics will do justice to the mega-combo-burrito filter.

    Finally, I think a lot of people forget that Schiit started or was at least instrumental in popularizing or bringing back the R2R or R2R/2R sound. The Yggdrasil, Gungnir, and Bifrost are hybrid resistor ladder (R2R) and resistor string (2R) DACs. Resistor strings are better in some ways in that their architecture guarantees monotonicity. The downside is that 2R architectures require 2^n parts so it's infeasible to go beyond a handful of bits.

    --

    The Modi MB is the odd hand-me-down bargain product with specific traits quite different from the Bifrost, Gungnir, and Yggdrasil. This is 2019 and things could certainly change in the future. The Modi MB actually reminds me more of the vintage R2R DACs than what I would call the "Schiit Multibit" sound. To me, there are "Schiit Multibit DACs" and then there is the "Schiit Modi Multibit".

    During the time R2R DACs were being phased out in favor of sigma-delta "S-D" DACs in the early-mid 90s, it was obvious that most S-D DACs were much more resolving than the R2R ladder DACs. Holdouts such as AG-D circa 2010+, or Wadia a few years before, used multiple BB PCM1704 chips in parallel to wring out as much resolution as possible, but even then, their efforts fell slightly short. Note that the BB PCM1704, which had a loyal following, kind of sounded like ass. PCM1702 less so. PCM63 before that was the pinnacle of the R2R ladder chips.

    The Schiit Multibit DACs (again, excluding the Modi Multibit) have their own sound. They do not totally sound like the R2R DACs of decades ago. And whether you believe in INL/DNL (think of this as amplitude jitter) over strict linearity measurements, the mil-spec tolerances of the chips used by Schiit in their DACs did contribute to their DACs edging out current S-D designs in terms of resolution. The latest S-D chips from AKM seem to have the best potential, but all other S-D chips from ESS (including the latest 9038), TI/Burr-Brown, Analog Devices, Cirrus, Wolfson, all seem to be at least a notch or two behind. Typically, the older the chip, the less resolving, all things being equal.

    During the time when Yggdrasil first came out, and this was years ago, almost every S-D DAC had fricking or odd nasty timbre to varying extents. It's too easy to forget this. Bottom line: Schiit Multibit DACs are extremely resolving of low-level information, more than even the latest S-D DACs, but they do this with a hint of that vintage R2R sound timbre.

    Basically, it comes down to four categories (NOS is an extra qualifier that needs to be discussed elsewhere) when it comes down to tone and timbre:

    Classic R2R
    UltraAnalog, PCM1704, PCM1702, discrete R2R ladders. I tend to detest everything here except for the UltraAnalog or Theta stuff, or the $5000+ Metrum models. Metrum, Wadia, Sonic Frontiers, Linn (their old stuff), CAL, A-GD, Airist (RDAC), Theta, Modi-Multibit and MOS16 (to a large extent, but not totally). Also Resolution Audio Cantata (which was not resolving).

    Reborn "R2R"
    Schiit, Holo (Spring 3), Soekris (these three are also very different from each other, but as a group is much more different than the above). Some discrete R2R ladder designs here, but they sound more modern and less syrupy or dark compared to the above.

    Bad Sigma-Delta
    ESS before their latest Sabre chips, badly implemented new ESS, badly implemented Wolfson, Cirrus (boring), BB (dull). This reflects the sad state of audio engineering. There's not enough money in it for the chip companies to care. Just think 5-10 years ago when most DACs sounded like ass: Mytek, Benchmark, etc. Others like Cambridge attempted to make things listenable by smearing everything with poo.

    Modern "good" Sigma-Delta
    AKM and Analog Devices, maybe ESS Sabre 9038, but need more time to tell. Newer implementations sound better. There seems to be more awareness on the part of designers of how S-D designs of the past exhibited unlistenable "digital" highs (rasp, glare, hardness, edginess, grain, etc.)

    Now coming back to the answer to the original question, here is the answer in bullet points:
    • Some people will want more of the classic R2R sound, that dense timbre, thicker bass, liquid mids, rolled highs, slower more romantic presentation, which the Schiit Multibit DACs (again, Modi Multibit excluded) only have to some extent. They will go elsewhere.
    • Some people dislike the closer placement of vocals or instruments mixed to be more upfront or the generally closer soundstage (row 10 seating) of Schiit DACs. They will go elsewhere.
    • Some people will complain that the Schiit Multibit DACs are too incisive and not relaxed enough. They will go elsewhere.
    • Some people will feel the Schiit Multibit DACs are not incisive enough. They will go elsewhere.
    • Some people will bitch that the Schiit Multibit DACs don't have a sufficiently "black" background. They will go elsewhere.
    • Some people will disregard the Schiit Multibit DACs because don't cost enough. They will go elsewhere.
    • Some people will dislike the irreverent "I don't give a fuck about what you think" attitude of Schiit. They will go elsewhere.
    • Some people will hate Schiit because they are popular. They will hate on them, bork measurements of their DACs, or present only information that suits their narratives.
    • Some people will move on because it's human nature to get bored.
     
    Last edited: Mar 19, 2019
  3. rlow

    rlow A happy woofer

    Friend
    Joined:
    Jul 18, 2017
    Likes Received:
    2,068
    Dislikes Received:
    3
    Trophy Points:
    93
    Location:
    Canada
    ^ Holy cow that’s great insight. Forget “sticky” this should just be posted on the homepage! ;)

    On a side note, I think that people moving on purely due to better measuring sigma-delta DACs should be called out seperately (with a note that says, show me what DAC measurements will actually tell you anything about how a modern DAC will actually sound, other than a totally garbage one). I realize the 2nd last bullet sort of speaks to this but not explicitly wrt S-D vs R2R measurements.
     
    dmckean44 likes this.
  4. geniekid

    geniekid Almost "Made"

    Contributor
    Joined:
    Mar 13, 2018
    Likes Received:
    140
    Dislikes Received:
    3
    Trophy Points:
    33
    Location:
    Chicago, IL
    I feel like there should be another bullet point - the reason I went elsewhere is because I don't have the ears/perceptiveness to hear the sonic differences mentioned previously and therefore prioritized on other features.
     
    Lyer25, pavi and atomicbob like this.

Share This Page