SMSL DL200 DAC & Headphone Amp Review

[jkim]

Thank you for a great review. I was looking for this long time.

Can you please comment on why this measurement was done just up to 10 kHz? Upper frequency region is considerably less audible for sure, but will you have possibility to provide this measurement up to 20 kHz?

Also do you have plans to review E1DA 9039S? It's such revolutionary small device.

In terms of what we can hear, response above 10 kHz in this test (THD+N vs. frequency) is just noise. We cannot hear any harmonic products excited by fundamentals above 10 kHz. Because these harmonics are beyond 20 kHz.

So, THD+N vs. frequency measurements above 10 kHz do not mean much. For the DL200, even if the response were rising a bit above 10 kHz, I wouldn't be concerned at all. Performance is outstanding regardless.

I would not review the E1DA 9039S. Its performance is no different than its prototype reviewed here.

 

[oleg87]

Performance looks solid. Is the screen OLED or LCD?

 

[nick_l44.1]

this measurement up to 20 kHz

I believe that the distortion of the 10000-20000 Hz frequencies is just as important. And the measurement bandwidth should be wide enough.
But I should note that such measurements are often not very accurate. Therefore, I additionally provide some accurate THD measurements at the end. The DL200 low gain was used for all measurements.
DL200 HPA BW190:

DL200 HPA BW190 32 Ohm:

9039s BW190:

9039s BW190 32 Ohm:

Accurate THD measurements taken with notch. Please, ignore noise characteristics and THD+N. With BW190 the total noise is of course very high.
DL200 HPA 1k (THD -130.8 dB):

DL200 HPA 1k 32 Ohm (THD -125.9 dB):

DL200 HPA 10k (THD -115.4 dB):

DL200 HPA 10k 32 Ohm (THD -111.8 dB):

9039s 1k (THD -139.3 dB):

9039s 1k 32 Ohm (THD -136.2 dB):

9039s 10k (THD -127.6 dB):

9039s 10k 32 Ohm (THD -116.6 dB)

 

[jkim]

I believe that the distortion of the 10000-20000 Hz frequencies is just as important.

This assertion can be misleading. Important for what? We need to clarify in what situation it may be correlated with potentially audible aspects of a device. I still consider harmonic distortion products (> 20 kHz) above 10 kHz fundamentals are generally less important in hi-fi audio applications.

 

[nick_l44.1]

This assertion can be misleading. Important for what? We need to clarify in what situation it may be correlated with potentially audible aspects of a device. I still consider harmonic distortion products (> 20 kHz) above 10 kHz fundamentals are generally less important in hi-fi audio applications.

Well, unless you consider extreme cases where THD exceeds -70-60 dB, no distortion can be heard directly, but still the sound is different. I myself can't hear loud signals above ~17kHz, but somehow I notice apodizing filter turned on at the wrong time. So, distortions of frequencies 10000-20000 Hz, although lying far beyond the limits of audibility, determine the interference of audible frequencies, and hence their purity and transparency. But to say that everyone should notice it at once is not possible. So many people are quite satisfied with the sound of chips like CS43131, although they have huge distortions at these frequencies. That's why there are objective measurements of distortion, and how much they are applicable to someone's feelings is another question. Just in my experience dl200 and 9039s have some relevant differences in sound clarity.
Also, as Bruno Putzeys said: "The practical upshot is that, strictly for the purposes of testing distortion and noise, the traditional “audio band” of 20kHz is a crude but useful stand-in for a human ear. The amplifier’s own passband, however, should continue well past that point."

 

[jkim]

I do think I have sensitive ears. But I cannot hear any difference between level-matched DACs that have reasonably transparent measurements.

I still believe objective measurements of devices for human hearing applications should be interpreted carefully along with scientific, psychoacoustic findings. In this regard, I totally agree to Amir's approach discussed, for instance, in this video. Here his talk is limited to the audibility of phase shifts, but the same principle should apply to all similar matters. Otherwise, we could easily fall for pseudo-science based on technical knowledge, like Paul McGown in the video.

 

[nick_l44.1]

I do think I have sensitive ears. But I cannot hear any difference between level-matched DACs that have reasonably transparent measurements.

I still believe objective measurements of devices for human hearing applications should be interpreted carefully along with scientific, psychoacoustic findings. In this regard, I totally agree to Amir's approach discussed, for instance, in this video. Here his talk is limited to the audibility of phase shifts, but the same principle should apply to all similar matters. Otherwise, we could easily fall for pseudo-science based on technical knowledge, like Paul McGown in the video.

There is no pseudoscience in the fact that the spectra of natural sounds are not limited by boundary frequencies or hearing thresholds. No matter what part of this spectrum can be perceived by the ear. Therefore, narrowing of these boundaries should be justified on the contrary. At the same time, application of the argument of inaudibility of too quiet or too high sounds is appropriate only for single tones and only statistically. The emergence of hires formats confirms the need for content of wider ranges.
I know of only one effect of not very high harmonic distortion. It's a loss of transparency.

 

[jkim]

There is no pseudoscience in the fact that the spectra of natural sounds are not limited by boundary frequencies or hearing thresholds. No matter what part of this spectrum can be perceived by the ear. Therefore, narrowing of these boundaries should be justified on the contrary.

So, basically what you want to measure is how well an audio device reproduces natural sound waves regardless of how humans process it. Right? That is exactly what we do not agree on. I do NOT deny what measurements show in the entire spectrum. But when we judge its implications and importance in hi-fi audio equipment, we should consider how we humans perceive it.

At the same time, application of the argument of inaudibility of too quiet or too high sounds is appropriate only for single tones and only statistically. The emergence of hires formats confirms the need for content of wider ranges.

I deem these arguments baseless. Sure, I understand what you mean. But its validity is very weak.

By the way, I am a statistical methodologist in a scientific field. Whenever human behavior/perception is involved, statistical analysis of data from controlled experiments is the gold standard in modern science. That is the main stream and dominating approach. Sure, one may want to object to it---I for one wish I could come up with a better approach. But it is what it is for now.

 

[nick_l44.1]

So, basically what you want to measure is how well an audio device reproduces natural sound waves regardless of how humans process it. Right? That is exactly what we do not agree on. I do NOT deny what measurements show in the entire spectrum. But when we judge its implications and importance in hi-fi audio equipment, we should consider how we humans perceive it.

Not sound waves, of course, but an electrical signal. Yes, that would be ideal and definitely unnecessary as well as unrealizable.
In fact, the justification for limiting the THD test to 10 kHz is for you to prove, not me.
I quoted Putzeys above that limiting the range to 20kHz is still quite crude. He also describes using the 18.5+19.5 kHz intermodulation test and compares it to a THD test to 20 kHz with no band limits.

The two tall poles 18.5 kHz and 19.5 kHz are the test tones. Even-order distortion components are visible at 1kHz and multiples, while odd-order components show up at 1kHz intervals from 17.5kHz down. So, despite using only audio-band signals and only looking at the audio-band outcome, this test is perfectly revealing of what an amplifier can do at the top of the audible range.

Even for linear amplifiers, I consider this intermodulation (IMD) test more meaningful than a non-bandlimited THD test with 20kHz.

For me this IMD test is less meaningful, simply because of my lack of experience. That's why I prefer to test THD up to 20kHz. But still I tried to do the described tests.
DL200 HPA. Pay attention to the 1kHz harmonic.:

DL200 HPA 32 Ohm. Good to know that the 1 kHz harmonic is almost the same:

9039s noload. Crazy -127.8 dB

9039s 32 Ohm. The 1 kHz harmonic is still invisible:

And CS43131 unloaded (100 kOhm actually) for compartion:

 

[jkim]

Not sound waves, of course, but an electrical signal. Yes, that would be ideal and definitely unnecessary as well as unrealizable.

Of course, I meant sound waves represented in electrical signals. Basically what you clearly indicated by stating

the spectra of natural sounds are not limited by boundary frequencies or hearing thresholds. No matter what part of this spectrum can be perceived by the ear. Therefore, narrowing of these boundaries should be justified on the contrary.

is just what I said:

what you want to measure is how well an audio device reproduces (or transmits) natural sound waves (or their electrically converted signals) regardless of how humans process it (when the signal is eventually converted to sound waves). Right? That is exactly what we do not agree on.

In fact, the justification for limiting the THD test to 10 kHz is for you to prove, not me.

In fact, it is not for either of us to prove. I already said harmonic products or noise > 20 kHz would be useful to examine in certain cases in which such measurements can be shown to be correlated with audible impacts. So, let's not digress by talking about other kinds of tests, which is not the point of this discussion. I have not seen a study in the literature suggesting a basis (or "just as important weight") for response > 10 kHz in this particular test of single-tone THD+N vs. frequency indicating potentially audible impacts. I do not see any such indication from the data that you posted in post #23.

I still have no problem with what I stated earlier:

I consider harmonic distortion products (> 20 kHz) above 10 kHz fundamentals are generally less important in hi-fi audio applications.

If we narrow down the topic to humans' ability to discern sound > 20 kHz---i.e., audibility of hi-res audio---, there have been some studies. An ASR member looked into the literature and summarized findings in this post. Sure, it is still an ongoing research topic. And we may not want to nitpick another person's potentially valid point, yours or mine. It's not productive at all.

EDIT. By the way, let's not continue discussing this topic in this thread. It is off-topic. I made a post asking a related question here.

 

[nick_l44.1]

I have not seen a study in the literature suggesting a basis (or "just as important weight") for response > 10 kHz in this particular test of single-tone THD+N vs. frequency indicating potentially audible impacts. I do not see any such indication from the data that you posted in post #23.

I don't see any point in testing THD+N vs frequency. In order not to miss distortion, the bandwidth has to be very wide, so in any case the total noise will not allow you to see the distortion pattern.
Except in the case of completely throwaway equipment, you will never see any audible level of distortion. So what audible influence can we talk about by looking at the fft? It will all depend on the amp conditions, the listener's hearing, and the particular music being played. But one can base some estimates on the differences in THD of the two devices. The DL200 has a little more distortion and it is possible to notice a slight difference by ear.
I wasn't disagreeing that not everyone will notice distortion above 10k. My disagreement is with limiting the frequencies tested to only 10k, and evaluating the quality of the device based on that data alone. Your statement is statistically correct. But according to the same statistics, there is a small but significant percentage of people who notice the differences. I believe this should be taken into account in the measurements.
I looked at the links, but that's not my topic. I'm more interested in how to “measure” the features of sound that I hear myself. And also what exactly I hear, in the sense of analyzing defects.

 

[xlinknz]

I'm new to DACs and looking to get one from SMSL, can anyone comment on the advantages of the DL200 over the newer DL100?

My use case is a DAC for a BR/CD/SACD transport connected via coax and playback of ripped SACDs (.DSF etc) via the DACs USB port

 

[JIW]

I'm new to DACs and looking to get one from SMSL, can anyone comment on the advantages of the DL200 over the newer DL100?

My use case is a DAC for a BR/CD/SACD transport connected via coax and playback of ripped SACDs (.DSF etc) via the DACs USB port

It has low and high gain making it better for use with sensitive IEMs and headphones. However, the volume control is just numbers instead of dB values and has a missing step. Also, some of the filters are mislabeled. Further, it has TRS instead of XLR connectors for the balanced line out.

 

[jkim]

I'm new to DACs and looking to get one from SMSL, can anyone comment on the advantages of the DL200 over the newer DL100?

My use case is a DAC for a BR/CD/SACD transport connected via coax and playback of ripped SACDs (.DSF etc) via the DACs USB port

One key difference is the DAC chips used in these two devices. The DL100 is based on CS43131 which is likely to produce the Cirrus hump that I reported here.

I'd choose the DL200. But if you don't need headphone output, how about the Topping D50 III?

 

[xlinknz]

@jkim @JIW thank you both for your replies

@jkim interesting the very comprehensive review you provided shows the CS43131 'bump' issue is dependant on the implementation i.e. some DACs that use the CS43131 don't have the issue and so far any testing of the DL100 for this issue is not published

I definitely want the option of headphone.

I also note the DL100 came out a year or so after the DL200 so possibly predictably has advantage of adding HDMI ARC but off set by lower cost DAC ship?