Hey everyone, I suspect many of you have already seen or heard of the Topping D10s by now [~US$100-110 at time of writing]. It has been out since spring/summer 2020, a DAC released during the early pandemic.
Basically, this is an update of the Topping D10 which I reviewed back in 2019with change to the DAC chip from the ESS ES9018K2M to the ES9038Q2Mwith improved specs - lower noise, higher dynamic range, etc.
I bought this through the usual retail channels as I'm planning to give the D10 away to a family member. Let's have a deeper look and consider the implications of this change in the DAC chip to the overall performance.
As you can see, the D10 and D10s look almost identical:
The only difference I see are the 4 front screw heads are of a different size between the two. The rear looks totally interchangeable. There's a slight difference in weight with the newer D10s being slightly heavier by 10g [D10 317g, D10s 328g - could literally just be the 4 larger front screws!].
Size is compact with the enclosure around 4" x 5.5" x 1.25" not counting the RCA plugs jutting out or small soft plastic feet.
Notice in the top picture that the manual contains some measurements like the D10 before. Good to see that the Chi-Fi companies like Topping and SMSL embrace the objective side rather than just vague advertising speak. There is a level of accountability when a company does this since the measurements can be verified by more sophisticated end users. The very competitive prices with high performance literally define superb value.
Feature wise, this is a basic DAC and identical to the D10. USB input only. Maximum samplerates up to 384kHz PCM [the usual family of 44.1 and 48kHz multiples] and DSD256. No separate power input needed. Gold plated analogue outputs, plus both TosLink and S/PDIF outs. No S/PDIF input, no headphone out, no switches to change stuff like filter settings, no remote. Like the D10, the front amber OLED display tells you sample rate and whether PCM or DSD is playing.
One other small difference between this and the D10 is that the opamp on the circuit board can now be replaced. The stock opamp in mine is the TI LME49720; an excellent low noise, low distortion part. I would love to see if any opamp rollers can find measurable improvements; years ago, I tried out the MUSES 02 opamp in the ASUS Xonar Essence One and did not notice much difference despite hoopla. My sense is that there is again a point of diminishing returns; good enough is good enough and putting in "better" parts will not necessary translate to audible improvements.
I opened the D10s for a quick peek one evening, here's the layout:
I. Objective Assessment
Let's have a look at how this puppy runs!
General measurement chain looks like this:
Raspberry Pi 3 B+ "Touch"with Volumio --> 6' shielded USB --> Topping D10s DAC --> 4' AmazonBasics RCA --> RME ADI-2 Pro FS ADC --> USB --> Intel NUC measurement computer
Looks good on the digital oscilloscope. Channel balance excellent with a peak Vrms of 1.95V [+8dBu]. This output level is a bit higher than the original D10 of ~1.5Vrms. Like the D10, Topping stayed with a linear phase digital filter:
Steep linear phase "orthodox" filter used. Notice that the impulse maintains "absolute" polarity unlike the D10 which inverted the signal.
This filter should work well in the frequency domain to suppress imaging artifacts as we can see using the "Digital Filter Composite" graph I've been using for years [based on Juergen Reis' suggestionsfrom back in 2013].
Yes, steep filter. In fact, steeper than the D10.
Okay, let's have a look at comparisons between this D10s with the D10 starting with standard CD-resolution a.k.a. 16/44.1:
Looking at the frequency response numbers and graphs, most notable is that the D10s has the steepest filter with what looks like a bit of passband rippling on the frequency response [I can confirm this finding using a REW sweep]. No worries, these are small variations less than +/-0.05dB below 15kHz [insignificant when we look at frequency response variation of speakers and headphones!]. The effect of the steep "brick wall" filter right around 20kHz can be seen on the D10s' noise and IMD+N sweep graphs as well.
Otherwise, notice just how "boring" standard 16/44 measurements look even with inexpensive DACs. While I still think it's useful to run 16/44 tests to make sure we're not missing something given that most of our digital music is still CD-resolution, it's rather unusual these days to run into significant issues with any reputable DAC.
Okay, moving on to hi-res, here's the 24/96 summary with graphs:
Okay, now we can differentiate among the DACs a bit better. We still see the fluctuation in the frequency response measurement of the D10s [again, small amount and now mostly ultrasonic]. The SMSL M100 Mk II's IMD+N anomaly is most obvious even though I doubt the vast majority of listeners would notice or complain [as discussed before].
Other than the stereo crosstalk being slightly higher than the D10[s], the DX3 Pro performed very well. The D10s proved all-around excellence in the measured parameters.
Okay, finally, let's quickly check out the 24/192 results:
| Only frequency response and noise level graphs. RightMark needs bug-fixing when displaying crosstalk and IMD+N graphs unfortunately. |
Again, excellent overall results on the D10s, the mild rippling ultrasonic irregularities in the frequency response not withstanding. Impressive how flat the old D10 frequency response is. Insignificant 60Hz hum with the D10s, excellent low noise. As usual, I'm not taking any special "audiophile" precautions like special cables or expensive power supplies; the D10s is simply connected to the Raspberry Pi streamer which itself is powered by a typical switching power supply.
DSD:
I almost never listen to DSD these days even though I still have a collection of stuff ripped from SACDs over the years on the hard drive [compressed with WavPack].
For completeness, for those who do play DSD or if you transcode PCM-to-DSD, here are some results to consider:
The DSD performance above are the results of transcoded RightMark test signals from 24/192 to DSD64/128/256 using JRiver 24, played back with the Topping D10s, and measured through the RME ADI-2 Pro FS.
The DSD results will not be as good as PCM since the PCM --> DSD conversion process will have very slight losses. Nonetheless, the results are clearly still "high res" quality.
Notice that the sweet spot here with the D10s is DSD128 [at least if you use JRiver's PCM-to-DSD conversion]. Here's a look at the frequency response and noise floor with these DSD measurements:
While DSD128 has rising ultrasonic noise above 40kHz, its overall noise level in the audible spectrum appears to be the lowest. DSD256 is still good but slightly more noisy [this is a typical finding].
Jitter time:
No issue with jitter. Most obviously, there is a pair of sidebands on the 24-bit test seen below -130dBFS and close to the 12kHz signal which makes any jitter distortion inaudible [as if jitter is ever really problematicin real life]. The 16-bit test looks good without gross level changes to the jitter modulation tone although we can make out a very small sideband pair at +/-4580Hz way down at -140dBFS.
For completeness, let's have a look at the 24-bit J-Test signal converted to DSD64 and DSD128:
Still see a few sidebands here and there but nothing worrisome, all below -130dBFS. Spurious ~6.6kHz noise in the DSD128 FFT.
THD+N:
Since it's commonly done, let's have a peek at the THD+N [SINAD] graph, 1kHz 0dBFS 24/96 sine fed to the DAC:
| Left channel shown. I also checked the Right channel which was similar with THD+N -110dB to make sure there was no gross disparity between the sides. |
As you can see, we have the detailed 1kHz signal with first 9 harmonics labeled for both the D10s and the older D10.
On my system, the D10s scores better than THD+N -110dB which is an objective improvement over the previous D10 -106dB result. As I have said in the past, when we get to these kinds of numbers, it's academic and I would not be concerned that this correlates to real-life listening differences. Nonetheless, it's nice to have more distortion-free dynamic range when we do stuff like DSP [as discussed here].
These are excellent results as one would expect with the latest generation ES9038 [D10s] family compared to the older ES9018 [D10]. The rest of your playback chain, not to mention the music recordings themselves would more than likely be noisier and way more distorted than this. ;-]
Frequency & Amplitude Stepped Sine:
Using Room EQ Wizard's stepped sine function, we can examine Harmonic Distortion vs. Frequency. In fact, we can have a look at the graphs across a number of output levels from 0dBFS to -12dBFS:
Notice that at 0dBFS, we see that the 3rd harmonic predominates through much of the frequency spectrum. As we attenuate the output level, the 2nd harmonic takes over as the largest contribution to the THD. Nice to see that the harmonic levels remain consistently low across the frequencies and there are no significant irregularities in any of the graphs.
Here's the Harmonic Distortion vs. Generator Level graph using a 950Hz signal from -120dB to 0dBFS. Notice the mostly linear "curve" with no unusual "hump". THD is lowest around -5dBFS [cursor]. Notice the 3rd harmonic rises above the 2nd from -3dBFS and up which correlates nicely with the Distortion vs. Frequency graphs that we just saw above.
There's a portion from -18 to -3dBFS where the 2nd harmonic features more prominently. Again, another academic observation - not that this would be audible at such low levels of course!
Using the graph above, we can have a look at thelinearityof the output level and see that it only consistently deviates by >1dB below around -110dBFS where the cursor is. This could just as well be the limitation of the measurement gear with unbalanced input. I don't think the human ears/mind will complain in any event!
II. Subjective
Well, as usual, I spent a couple of nights mainly listening to the Topping D10s as a desktop DAC with the Emotiva Airmotiv B1+ speakers [SMSL SA300 Class D amp] and through the Drop + THX AAA 789 headphone amp with modded Dekoni Blue mainly, with some time spent on the Sennheiser HD800 as well.
So how does this sound?
It's excellent if you're into high-fidelity ;-]. Francesca Dego's performance on Il Cannone: Francesca Dego Plays Paganini's Violin[2021, DR14] sounded fantastic with beautifully rich harmonic texture. Superb recording that sounds realistic without synthetic shrillness. By the way, there's a great section in the liner notes about the history and significance of Paganini's "Il Cannone" violin by the curator of the instrument in Genoa - I was there a couple years back, what a beautiful city!
I hadn't heard it before, but recently I checked out one of Patricia Barber's older albums Split[1997, DR14]. Wonderfully propulsive pacing on "Early Autumn", nice piano work on "Retrograde", and the sappy love song "Then I'll Be Tired of You" showed off Barber's unique, at times dramatic vocals. I will need to revisit this album a bit more in the days ahead for enjoyment rather than trying to be critical about sound quality!
For a little bit of an '80s kick, I had a listen to Kajagoogoo & Limahl's Too Shy - The Singles and More[1993, DR13]. Not the best audio recording but it's great to enjoy oldies like these again, hearing them knowing the "accuracy" of the audio components are way higher thanI would have ever heard them before. Listening to "Too Shy" or "Never Ending Story [Club Mix]" brought back memories and even though the instrumentation sounds dated and "thin", lacking in deep bass by today's standards, the sound is as I remembered and the DAC delivers the sound from the source recording without editorializing or coloration.
Finally, this DAC has no problems with rendering bass frequencies which are plentiful on Yello's latest album Point[2020, DR10]. Have a listen to "Siren Singing" or "Zephyr Calling". As usual, this is "synthpop" type stuff with lots of quirky sounds, artificial "surround" effects, weird lyrics [check out "The Vanishing of Peter Strong"], and generally just fun. Great test material for the times when a visitor comes over and you want to make an impression. For this album I also had a listen with the Topping D10s connected to my main system downstairs with subwoofers going. Nice expanded soundstage, and the illusion of "surround" from 2-channels. I see that this album is also available as a Blu-ray Dolby Atmos Edition which should be enjoyable [I have not heard it yet].
III. Summary
| What DSD256 [11.2MHz, 1-bit] playback looks like. |
Well, if you're in the market for a straightforward low power [ TosLink/Coax on a TV or receiver.
The only criticism I would make is with the frequency response rippling. In audio as in life, there's always a balance to strike within the limits of a device or opportunity in life. As much of a proponent as I am for sharp linear phase filtering, there is such a thing as going "too far"! That steep 20kHz filter is unnecessary. I would have preferred if Topping had just used the previous D10 parameters without the upper frequency undulations. I wonder if a firmware update can address this.
Otherwise, nice DAC performance update, Topping. Good execution on an accurate, no-frills DAC which anyone can afford. In fact, given the level of performance, I think many audiophiles will be surprised at how good this sounds compared to much much more expensive devices.
On a side note, IMO this is the kind of low distortion performance that I expected the AudioQuest Dragonfly Cobalt[same DAC chip]should have had. Instead, in all their "wisdom", I think AQ's engineers employed misguided beliefs about filtering plus the so-called benefits of MQA thus handicapped that product's potential. I still remember feelingdisappointment when I reviewed it back in 2019. To other audiophile DAC companies: please don't be like this, and make sure not to sell products like the JitterBug unless you're able to publish data to justifywhy there's a need for such a thing! Needless to say, I'm very weary of questionable companies like AudioQuest in the audiophile space.
--------------------
| Aligned spectra from 45 seconds music recordings from two modern, excellent fidelity, DACs of very different price usingDeltaWave.Notice one of the DACs [white] has earlier and steeper filter than the other [blue]. Nice example of perceptibly perfect match for human listening IMO. |
Are we at "perceptibly perfect" with even $100 DACs?
Admittedly as a "reviewer", objective measurements and subjective "listening to" DACs have generally become desperate exercises in trying to findminisculefaults over the years.I'm sure you know all about "bit-perfect", and over the last decade,I believe we have arrived at a place where if we define sound quality as "high fidelity", we can easily find "perceptibly perfect" DACs capable of resolution beyond the needs of human hearing.To be concrete, what I mean is that if we take devices meant to reproduce digital audio dataaccurately, within the context of a volume-controlled, "blind" test where someone is unaware of what DAC is being used, I believe humans would not be able to differentiate between an accurate $100 DAC like this little Topping D10s and an accurate $5000+ DAC [take your pick].
[In fact, the other day I brought the little D10s over to a friend's place to compare with his very expensive DAC in an informal "shoot-out". Let's just say he does not want me to talk about the result. ;-]
I know...Manyaudiophiles will disagree with the above, but I don't think many have subjected themselves to controlled testing either to know for sure. [If you recall, a few years ago we even did a little blind test here using recorded music from different devices.]
Subjective-only reviewers [and perhaps certain manufacturers] will not want to hear this because it implies that one can no longer use subjective descriptions of sound quality to differentiate many modern products meaningfully. Such descriptions of supposed sound differences between high fidelity DACs become exercises in creative writing and imagination, not factual discriminations. This is a dangerous meme because it means audiophile magazine editors will also lose the ability to declare "Best Sounding DAC Ever?" as their lead article every few months! :-] In fact, this kind of talk also tends to get threads on discussion forums closed or even worse erased [so much for free speech], which adds to the bias against honest discussions on the Internet. This is why I prefer expressing my opinions in my own blog.
I'm of the opinion that the reviews ofmeasurement-confirmed accurate DACsare most useful when the writer pairs objective information with discussion of unique features, build quality, usability, even esthetics rather than spending too much time describing sonic attributes which are unlikely to translate meaningfully anyways. Since I believe subjective-only claims are unreliable with good modern DACs, I'm particularly mindful about the reviewer falsely creating reader biases which I think also does a disservice for manufacturers that actually deserve to be recognized for producing high-value products. Companies like Topping or SMSL I doubt will ever be reviewed in the pages of The Absolute Sound orStereophile yet as far as I'm concerned, sound just as good as the fancy brand names they feature, and definitely deserve more consideration than insanely priced devices or even the aforementioned AudioQuest Dragonflies which IMO are not all that special.
Assuming that a reviewer normally listens to an "accurate" reference DAC, in the event where he/she clearly hears a difference with another product, it's just as likely that this is due to inferior performance - something that sounds "different", even if "preferred" does not mean "better" fidelity. Some listeners will have a preference towards "inaccuracy" just like some folks will prefer EQ with bass and treble pushed up to levels which are clearly no longer "neutral" [that's OK if you know what you're doing and not uninsightfully claiming to everyone that this is how it should sound!]. Likewise, in a more subtle way, the amount of distortion that a device produces could be euphonic for some listeners asdiscussed recently with the Pass SIT-2 amp[also OK if you know what you're listening to]. The problem is that subjective-only reviews can't actually tell us what's going on most of the time!
As audiophiles, I think we need to think and talk about these things not purposely to stir controversy in some circles, but rather it's important to be honest and truthful to ourselves and in the service of the greater High Fidelity Audiophile Hobby. If we care about high fidelity as a primary objective, then we need to make sure that recommendations made are not simply the results of biases built on perceived luxury, or just advertising dollars at work simply in the service of consumerism.
Enough for now...
Take care everyone. I hope you're enjoying the music!