Saturday, 15 June 2019

MEASUREMENTS: Topping D10 DAC. (And a few words on "absolute polarity / phase".)

Hey guys, getting busy around here as I'm preparing for summer holidays coming up in a couple of weeks :-).

I did want to post a "quick" report however on the Topping D10 DAC (<US$90) I got last week... It's for an upcoming project of sorts which I'll post on over the months ahead. What I wanted was a DAC that could be powered off USB, reliable with Window and Linux compatibility, that's reasonably portable, and of course of high signal accuracy.

Notice the DAC manual shows some AP measurement graphs... I guess Topping believes in showing objective accuracy :-). Nice.
As some of you will likely already know, this DAC was measured by Amir at Audio Science Review initially back in the Spring of 2018 using his Audio Precision gear and more recently updated with March 2019 results here. Looks good. I did some measurements to confirm good performance with the unit I received so the results here will hopefully touch on a few items I haven't seen and show a few of the tests I typically perform here on the blog that may not be shown elsewhere...

For the tests here, I'll just be using my previous Focusrite Forte ADC. In part this is because for some upcoming testing, plus the portability of the Focusrite Forte is important even though it's not as capable as the RME ADI-2 Pro FS I've been using over much of this past year.


The Topping is a small, but relatively handsome all black DAC with sturdy metal case, and featuring a good sized front LED. It's capable of up to 32/384 PCM and DSD256 playback. The DAC is based on ESS Tech's ES9018K2M chip and XMOS XU208 USB2.0 interface. On the back are the USB-B input which also powers the machine, both coaxial and TosLink S/PDIF digital outputs are available, plus the unbalanced analogue RCA output of course. I have not used the digital outputs yet so can't say if both the coaxial and TosLink can both handle 24/192 (most modern DACs these days can manage TosLink 24/192 in my experience), nor examined the jitter out of these interfaces.


I've been running the D10 fed with my Raspberry Pi 3 using piCorePlayer streaming off Logitech Media Server. Setup was straight forward and no compatibility issues thus far, and I've streamed up to 24/384 without issue. The measurement chain looks like this:
Raspberry Pi 3 (piCorePlayer), WiFi, 5V lithium battery powered --> shielded USB --> Topping D10 DAC --> shielded 6' RCA --> Focusrite Forte ADC --> Win 10 laptop (battery powered)
As you can see, the whole measurement chain is battery powered. The Focusrite Forte ADC is run off the laptop itself. The Raspberry Pi 3 and Topping D10 are both powered off a RAVPower 22000mAh lithium power bank which can easily run the unit for hours on end.

Let's see some signals through a digital oscilloscope:

1kHz 0dBFS sine and 1kHz -3dBFS non-ringing square 24/44.1.
It looks pretty good. Nice right-left channel balance. RMS output of 1.5V from the RCA connectors. Square wave suggests symmetrical linear phase digital filter. Indeed, that is what it is:


Notice that the impulse is inverted. Not a problem unless you're a strict adherent of "absolute polarity". We'll talk more about this later.

Let's overlay a few FFT graphs to look at that filter in more detail (this is the "Digital Filter Composite" [DFC] graph I often show on this blog site, based on Juergen Reis' work)...


What this tells us is that the filter can handle intersample overload even with the 0dBFS wideband white noise (this generally means it can handle +3.01dBTP material just fine). Filtering is also reasonably steep with essentially no imaging distortions from the 19 and 20kHz tones. Some harmonic distortion but at least 90dB below the primary signals in the ultrasonic frequencies. This is a good looking DFC graph for a DAC's built-in filter.

Okay, let's now have a look at the RightMark battery of tests with some comparative results...

24/96:
Note: Although not in the label, the TEAC UD-501 was using XLR output.
Note: TEAC UD-501 using XLR output.


24/192:



Overall the Topping D10 DAC compares favourably to the other devices, including some much more expensive DACs like the old Oppo BDP-105 and my TEAC UD-501. Notice that the noise level of the Topping was subtly higher than the others but what is impressive are the excellent THD and IMD results. In this field of comparison devices, only the TEAC UD-501 24/96 with XLR outputs approached the THD result from the Topping. Obviously if I measured my newer and better DACs like the Oppo UDP-205 and RME ADI-2 Pro FS with balanced output, these would be superior.

The two less expensive DACs - the HiFiBerry DAC+ Pro (HAT board) and SMSL iDEA performed significantly lower in regards to distortion even though noise level from these DACs were slightly lower.

The low distortion results are exactly what I needed for the measurements I wanted to perform which will be discussed in the near future.

I ran into troubles trying to plot the frequency response of the different devices at 192kHz so I just showed the frequency response of the Topping D10 alone instead. We see that the high frequency extension goes out to ~72kHz by the -3dB point, and about 83kHz at -6dB; of course at these ultrasonic frequencies, the Focusrite Forte ADC itself could be the limiting factor. Comparative noise levels could still be plotted without issue with device overlays, again demonstrating the slightly higher noise floor of the Topping.

Finally, here are the J-Test graphs:


While we see some low-level jitter in the 16-bit result below the jitter modulation tone, it is in the 24-bit J-Test that they're more evident as symmetrical sidebands. Note though that these are very low amplitude products - something like 130dB below the primary signal.

Jitter's interesting when found, but these are trivial amounts and nothing close to being audible! (Remember to put the whole jitter "issue" into context by examining and listening to some simulated jitter samples.)

In conclusion...
For <US$100, one can get some very capable high-fidelity DACs these days from the various "Chi-Fi" brands like Topping or SMSL. No surprise given that most high tech manufacturing comes from China these days including your latest and greatest iPhones et al...

The Topping D10 DAC sounds clean and tight when I hooked it up to the main system over the last few evenings (Emotiva XSP-1 preamp, Emotiva XPA-1L monoblocks, Paradigm Reference S8 speakers) for some music listening. Though I heard that the movie wasn't all that good, the X-Men: Dark Phoenix soundtrack from Hans Zimmer is highly reminiscent of the Interstellar soundtrack if you like that kind of hard driving, bass-filled, aggressive sort of sonic affair! Unfortunately this "kind" of sound also generally means dynamic range compression these days and the album shows a DR9 result.

Speaking of soundtrack music, I recently was reminded of the Finding Forrester (DR11) and had a listen using this DAC. A nice sounding sampling of jazz including "Iz" Kamakawiwo'ole's version of "Over the Rainbow / What a Wonderful World", plus a selection of Miles Davis for those who want a convenient mix playlist on one CD... Nice separation of instruments, good soundstage through the Topping D10, I can't complain.

I also had a listen to Morgan James Live from Dizzy's Club Coca-Cola: A Celebration of Nina Simone (DR9). Good sounding live female vocals - very audiophile-friendly :-). Lots of detail off this DAC with nice ambiance from the live setting coming through. Good tonal reproduction without any unexpected or excessive sibilance. The little noises and clapping from the audience members sound detailed and well placed.

As you can see from the measurements, this DAC performs with very low distortion performance. It might give up a dB or two in the noise-floor and stereo crosstalk departments compared to more expensive devices, but otherwise IMO you're not missing anything. Notice that the DAC has no buttons or any kind of hardware configuration. It's simply a no-nonsense asynchronous USB DAC providing good single-ended RCA output that delivers technically accurate performance down to the textbook linear phase digital filter (my preferred setting), and will cover a wide range of PCM and DSD samplerates - likely more than we'll ever need. No nonsense like MQA of course.

The only thing unusual about the DAC was the inverted polarity (or "phase") mentioned above. While I would have preferred that the polarity were not inverted (ie. a positive deviation in the audio editor resulted in the same positive deviation at the DAC output), this isn't really an issue IMO. In fact, if you're a computer audio user, it's quite easy to "fix" this; a number of the popular programs can do it... Here are JRiver and Roon's settings for example:

Can select that polarity can be inverted depending on the channel... Make sure to invert polarity for both left and right of course.


I know that the significance of "absolute polarity" can be a bit contentious with some audiophiles claiming they can hear it and others can't. I've tried over the years and cannot meaningfully hear a difference with various headphones and good quality speaker systems. While you can read opinions from various audiophiles, make sure to also check out this forum thread for discussions from the pro/studio audio folks. Remember that in the albums we buy, there is no assurance that a studio will follow a standard in the music one way or the other.

Let's think about this. Remember that sounds are waves; the impulse response I use to easily demonstrate polarity is actually an "illegal" signal and not an actual sound one should ever "hear" in a recording. In order for us to actually hear waves at any given frequency, we need to experience both the compression and rarefaction for a period of time in order for the brain to experience the event. Does it really matter if that very first motion of one's speaker/headphone transducer pushes the air forward or pulls the air backwards when one actually needs a few of those vibrations to register the "sound" experience? I suppose in the very low frequencies, the appropriate initial "push" of air from a deep bass drum might be experienced as more "right" although I personally have not experimented with this. [UPDATE: See the next blog post on "absolute polarity" discussing and clarifying comments...]

I think J. G. Holt's assessment back in 1980 "Absolute Phase: Fact or Fallacy?" is worth reading. As explained, likely the most significant difference that may be heard when flipping between polarity settings is when the transducer (or even the amplifier) is nonlinear and functions asymmetrically. The asymmetry would result in playback distortion; this is bad and suggests a playback system issue. If so, it's then important to keep in mind that if one were to hear a big difference "flipping the polarity switch", could it be that the amplifier/transducers you're listening to may be of poor quality? (Rather than the idea that one has found the "right" setting for the music itself...)

Remember that multi-way speaker systems often have drivers set up in opposite polarity, so the idea of an "absolute" here might not even exist for one's system and it can depend on the frequency of the sound itself which "way" the air is being pushed.

Anyhow, since I guess I can't generally seem to hear this polarity issue and/or the speakers I use are "linear enough", there's nothing I can do to explore/test this further. I suppose we could set up some kind of Internet Blind Test and see if in a blinded state, the audiophile community out there can come to consensus with a few demo tracks of which sounds "better"...

I'd love to see any research references on the "audibility" of absolute polarity. Drop links / references in the comments below if you know of some...

Regardless, for my purposes which will become evident ahead, the Topping D10 sounds very good and work just dandy. :-)

[***Note: for those who bought the D10 awhile back, make sure to update to the newest firmware v1.02 (18l) for some bug fixes.]

Have a great week ahead... Hope you're all enjoying the music as we head into the summer holiday season here in the Northern Hemisphere.

24 comments:

  1. Amir has tested a number of Topping DACs, one of which is often used as a standard to compare other DACs against. Topping DACs almost always reflect state of the art performance. (I owned a Topping D20 for a while. It was excellent in that it sounded just like...a DAC.) Topping performance demonstrates how ludicrous it is to pay megabucks when you can get identical sound for around a hundred bucks or so.

    As for polarity, if it really matters to someone, switch the connections on your speakers and voila! Correct polarity for no additional cost. Not that you could tell the difference in the first place.

    For anyone wishing to read what polarity obsession can do to a man, read Clark Johnsen's polemic, and Peter Aczel's hilarious response, here: http://www.biline.ca/audio_critic/mags/The_Audio_Critic_18_r.pdf (Aczel calls it "the longest crank letter ever published". If it's not, it's damned close.

    ReplyDelete
    Replies
    1. Zut alors! How could I have missed such an epic "Letter to the Editor" after all these years. :-)

      Thanks for that link to the Audio Critic issue. I started reading page 3 and got tired so skimmed the rest of it. Hilarious pseudo-intellectual, pompous, and pretentious bloviation!

      Delete
    2. Aczel's vehement response is pretty fun reading, too.

      Delete
  2. Up to the two years ago, I thought absolute polarity is not important. Then I performed absolute polarity listening test and found the sound becomes very different for some waveform by just invert polarity. This is the my test result with two test FLAC files: https://sourceforge.net/p/playpcmwin/wiki/AbsolutePolarityListeningTest/

    ReplyDelete
    Replies
    1. Wow.

      Good show Yamamoto-san! Indeed your test files were easy to ABX on my Sony MDR-V6 headphones and ASUS Xonar Essence One DAC's headphone output...

      Here's my ABX log:

      foo_abx 2.0.6c report
      foobar2000 v1.4.1
      2019-06-15 21:10:58

      File A: AbsolutePolarityListeningTestTone.flac
      SHA1: c4f4e59eedb9e838e5c393c4508e1f29113591f0
      File B: AbsolutePolarityListeningTestToneInverted.flac
      SHA1: d093c9a9723042caf0b39d1afa8d814660100808

      Output:
      DSD : WASAPI (push) : Speakers (ASUS Xonar Essence One), 24-bit
      Crossfading: NO

      21:10:58 : Test started.
      21:11:10 : 01/01
      21:11:21 : 02/02
      21:11:31 : 03/03
      21:11:40 : 04/04
      21:11:50 : 05/05
      21:12:00 : 06/06
      21:12:09 : 07/07
      21:12:17 : 08/08
      21:12:25 : 09/09
      21:12:32 : 10/10
      21:12:43 : 11/11
      21:12:53 : 12/12
      21:13:01 : 13/13
      21:13:10 : 14/14
      21:13:20 : 15/15
      21:13:28 : 16/16
      21:13:28 : Test finished.

      ----------
      Total: 16/16
      p-value: 0 (0%)

      -- signature --
      9bdb7ac25c2d9e1be17598806f81114426e5d48e

      Easily heard and done in less than 3 minutes with 16 trials. Which then leaves me with the question of whether my Sony V6 headphones might have asymmetrical playback issues or is this direct evidence that our hearing is capable of distinguishing "absolute phase"!

      Will have to pull out some of my other headphones as well and have a listen...

      Given that this is a synthetic test signal, I wonder is there now evidence to show this discernible effect in music signals? Anyone have a snip of an actual recording where polarity is easily heard?

      Delete
    2. Easily done with AKG Q701 and even easier with Sennheiser HD800...

      Sennheiser HD800 log:

      foo_abx 2.0.6c report
      foobar2000 v1.4.1
      2019-06-15 21:28:19

      File A: AbsolutePolarityListeningTestTone.flac
      SHA1: c4f4e59eedb9e838e5c393c4508e1f29113591f0
      File B: AbsolutePolarityListeningTestToneInverted.flac
      SHA1: d093c9a9723042caf0b39d1afa8d814660100808

      Output:
      DSD : WASAPI (event) : Speakers (ASUS Xonar Essence One), 24-bit
      Crossfading: NO

      21:28:19 : Test started.
      21:28:29 : 01/01
      21:28:36 : 02/02
      21:28:42 : 03/03
      21:28:47 : 04/04
      21:28:53 : 05/05
      21:28:58 : 06/06
      21:29:03 : 07/07
      21:29:09 : 08/08
      21:29:15 : 09/09
      21:29:20 : 10/10
      21:29:25 : 11/11
      21:29:31 : 12/12
      21:29:37 : 13/13
      21:29:46 : 14/14
      21:29:51 : 15/15
      21:29:56 : 16/16
      21:29:56 : Test finished.

      ----------
      Total: 16/16
      p-value: 0 (0%)

      -- signature --
      994dc2ea717ee73cd8b9ad7b05005def8970cc52

      This is certainly interesting and am happy to change my mind on this one once I check things out a bit more :-).

      Delete
    3. A quick update on the listening test. I just tried the test signal using the headphone output of my Oppo UDP-205 with the Sennheiser HD800. I've found that it was much more difficult with the Oppo than through the ASUS Xonar Essence One on initial listen. Haven't tried ABX yet.

      Since I believe the Oppo is a better playback device, it's quite possible that what I am hearing is a result of the "asymmetry" in reproduction as a result of the DAC/headphone amp. Hmmm, could it be that as I suggested, more accurate devices should result in less audible polarity effect?

      Delete
    4. Hi Archimag-san,

      Thanks for testing, I created this tone to mimic trombone waveform so solo trombone music may be easy to distinguish absolute polarity.

      And I tested several trombone albums with polarity switch and found it is very difficult :)

      Delete
    5. Nice! I'll post more on this with the next blog entry next week.

      This is very interesting as I (like Gilles below) try out a few headphones and DACs in my collection!

      Delete
  3. There is a big contrast between you and that pink panther reviewer. Your style is so calm and I like your style.

    ReplyDelete
    Replies
    1. Appreciate the note Bennet.

      Yeah, it's good to be calm and evenly keeled. Enough bluster and anger in this world already. No need to be unnecessarily rude...

      Of course, it's important to still be able to call BS on that which is and say a thing or two to those in the audiophile world who should know better :-).

      Delete
  4. Yes, the test tones do sound different with the 3 headphones I tried. I found some short solo trombone music that might make a good real music comparison. I see why the trombone was chosen. The waveform is very non symmetrical...inverting it looks very different. I think I hear a very slight difference but if anybody would like to do a proper abx test, here are the files (12 sec. flac files from a 16/44.1 wav capture of a You Tube video)

    https://sites.google.com/site/ptqspecprof/temp/trombone.flac
    https://sites.google.com/site/ptqspecprof/temp/trombone_inv.flac

    ReplyDelete
  5. A small note. I can't seem to place my comment at the proper place...even if I click on the indented Reply button, my comment always end up published at the end. Don't know why but it's annoying.

    ReplyDelete
    Replies
    1. Thanks for the trombone samples! I'll have a look and listen. Fascinating :-)

      Don't know why the indentation is not working for you. Generally I just click on the "Reply" and it'll do its thing.

      Delete
    2. Seems I have to be signed in BEFORE I write the comment! It's ok now.

      Delete
    3. By the way, the inverted file I marked _inv seems to be the one with correct absolute phase since I included at the start a portion of the silence preceding the emission of sound and on this file the first cycle goes up while it goes down on the other. I guess the phase accuracy of YouTube (or the recording) can't be trusted. I do think the _inv file sounds slightly better with sharper attack and a somewhat smoother sustain, but it may be subjective...

      Delete
    4. I had a look in the audio editor Gilles, yes, indeed it looks like the _inv trombone file is the "absolute polarity" one with the asymmetry more positive...

      Delete
  6. Folks interested in the audibility of 'absolute phase' are referred to Audio Critic issues 18 and 22 letters section (on-line with a search). Another commentor cites it, but I'd like to expand. In the early '90s a man named Clark Johnson was citing something called the "Wood Effect", arguing that absolute phase was not only audible, but a strong determinant in reproduced sound quality. He wrote a book about it, I think. Of course he did not 'prove' his claim using DBT, but made an assertion, and came up with a 'new' test protocol that, in his opinion, could be used to demonstrate phase audibility to anyone's satisfaction.

    Corey Johnson (professor of experimental psychology) critiqued Johnson's "triple blind test protocol" as nonsense, and R.A. Greiner (professor emeritus EE) wrote in to discuss his recent JAES paper on polarity. Dr. Griener's experiments showed:

    1) acoustic polarity is audible with steady state (monotone) signals.

    2) polarity with simple musical passages was 'extremely difficult' to ascertain in a 'highly idealized simplified laboratory environment'.

    3) on complex musical passages in a normal living room no subject was able to reliably determine phase differences.

    Dr Greiner stated that he discussed the issue with both Drs (Richard C) Heyser and (Stanley) Lipshitz at AES meetings. Their joint conclusion was that while it would be ideal if the recording industry adhered to a standard, in the scheme of recording problems absolute phase is pretty low, to not a concern at all, in the list of special effects that need addressing.

    ReplyDelete
    Replies
    1. PS: It was Jeff Corey (not Corey Johnson) who critiqued Clark Johnson. My brain's absolute polarity was confused. I apologize for not getting it correct.

      Delete
    2. Thank you.

      Perfect... I've been looking into this last couple days and seems to be exactly what the tests and listening have been suggesting. Cool stuff out of the biology/neuroscience literature as well about this.

      Will post more on the weekend!

      Delete
  7. Hi Archimago,

    I sent you direct messages on Twitter a week ago, I think you night have missed them because you didn't reply. Can you please take a look? Thanks!

    ReplyDelete
    Replies
    1. Hi Otto!

      Got the message. Sent a reply and will connect soon...

      Delete