• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

SMSL D400ES DAC Review

Rate this DAC:

  • 1. Poor (headless panther)

    Votes: 12 5.1%
  • 2. Not terrible (postman panther)

    Votes: 11 4.7%
  • 3. Fine (happy panther)

    Votes: 58 24.8%
  • 4. Great (golfing panther)

    Votes: 153 65.4%

  • Total voters
    234

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
I'm dumb so explain this to me like I'm a 5 year old... Why not just add a negative gain of a few dB in software? Shouldn't that avoid clipping?

Use the dac in fixed mode , software dsp and no Filters. Makes more sense to EQ yourself anyway instead of using some flavour like filters

That solution does not apply. It always goes max in the software. Otherwise, you immediately lose quality and dynamics. Not to mention that if you use kernel streaming or Tidal mqa (exclusive mode) you can't even lower the volume even though you want to for some reason. I agree with you about the EQ, but they cannot be applied to mqa, and they are in the vast majority with Tidal. And they really sound better than PCM 44.100 (CD).
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
...and not only SMSL
Sabaj%20A20d%202022%20-%20DFC%20Filters%201-4.png



Surprised? But I am a long time Archimago blog reader, so no surprise for me.
Filters%2B3%2Bonly.png


Even Lynx...
DFC.png


I don't use MQA, but if it is the case, then the DAC / playback software should automatically calibrate the filter level when playing MQA content so that the output doesn't clip. It is MQA's responsibility to handle these things.
 
Last edited:

Snoopy

Major Contributor
Joined
Jul 19, 2021
Messages
1,628
Likes
1,206
That solution does not apply. It always goes max in the software. Otherwise, you immediately lose quality and dynamics. Not to mention that if you use kernel streaming or Tidal mqa (exclusive mode) you can't even lower the volume even though you want to for some reason. I agree with you about the EQ, but they cannot be applied to mqa, and they are in the vast majority with Tidal. And they really sound better than PCM 44.100 (CD).
You don't lose anything with digital volume control and a negative gain if it's in 64 bit like roon does. It's not the same as adjusting the volume in windows:)
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
What I am thinking is the D400ES cannot deliver more than 4.2Vrms regardless of bypassing the filter or not. Because the filter and frequency response measurements are often not done in anywhere close to 0dBFS, there is actually a chance that the D400ES will clip when using 0dBFS 1kHz sine output when the filter is bypassed. Whether this is true or not, or the cause is due to different product revision or not can only be confirmed by measuring both the D400ES units to confirm, and absolutely not by listening.
 

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
You don't lose anything with digital volume control and a negative gain if it's in 64 bit like roon does. It's not the same as adjusting the volume in windows:)
You're right, but even if you do that, you lose clarity and dynamics that you can't fully make it up by turning up the volume on the amplifier. But it is much better than when the filter is turned on. Then it looks like a curtain has been drawn over the speakers. So even though you turn up the volume on the amplifier, it seems as if the speaker is louder under the curtain. A variant with volume reduction on the DAC is the best compromise. After correspondence with SMSL, that's it.
 

Trell

Major Contributor
Joined
May 13, 2021
Messages
2,752
Likes
3,285
You're right, but even if you do that, you lose clarity and dynamics that you can't fully make it up by turning up the volume on the amplifier. But it is much better than when the filter is turned on. Then it looks like a curtain has been drawn over the speakers. So even though you turn up the volume on the amplifier, it seems as if the speaker is louder under the curtain. A variant with volume reduction on the DAC is the best compromise. After correspondence with SMSL, that's it.
Are you able to hear a few dB decrease in volume from Windows causing audible noise or lack of clarity or dynamics with this DAC? I doubt that is the case.
 

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
Are you able to hear a few dB decrease in volume from Windows causing audible noise or lack of clarity or dynamics with this DAC? I doubt that is the case.

I listen to music either through Tidal directly in MQA (exclusive mode), or through Audirvana (same mode) along with my collection. Kernel streaming, or Wasapi. So, there is no volume manipulation. Of course, during the month and a half that I've been here, I've tried everything, including turning off the exclusive mode and reducing the volume through the program. Definitely the best compromise is to lower the volume on the DAC. Unfortunately, the filters degrade the sound a lot, so they are OFF That's how I listen.
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
The better perceived clarity or dynamics could be caused by clipping, this is how loudness war mastering works. When people get used to hear such kinds of sounds, the non-clipping version could indeed sound worse to some people due to subjective preference.

A lot of studio interfaces can deliver more than 4Vrms without clipping so it is not necessary to limit yourself to the MQA models. If you already own more than one MQA DACs, I don't see why you cannot get a non-MQA one with higher non-clipping output for non-MQA content, unless you almost listen to MQA exclusively.
 

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
6,874
Likes
6,672
Location
UK
You don't lose anything with digital volume control and a negative gain if it's in 64 bit like roon does. It's not the same as adjusting the volume in windows:)
You're right, but even if you do that, you lose clarity and dynamics that you can't fully make it up by turning up the volume on the amplifier. But it is much better than when the filter is turned on. Then it looks like a curtain has been drawn over the speakers. So even though you turn up the volume on the amplifier, it seems as if the speaker is louder under the curtain. A variant with volume reduction on the DAC is the best compromise. After correspondence with SMSL, that's it.
The only real thing that could happen when you use 64bit and reduce volume digitally is that you will lower the SNR (Signal to Noise Ratio) - this just means that if you lower it too much then you might start hearing hiss when music is not playing (very quiet or silent sections of the track). If you don't hear hiss then you're fine, you're not losing anything. 64bit or 32bit or 16bit, you still lose SNR when you use digital volume control. But using 32bit & 64bit (& 24bit if you don't reduce volume as much) will allow you to retain all those "volume steppings" that are included within your 16bit recorded music - which I think is described as quantisation error, so using (24bit)/32bit/64bit then you're protecting yourself from quantisation error and retaining all the "volume steppings" within your 16bit recorded music when using digital volume control. That's as much detail as I understand it. Basically, digital volume control is fine & nothing to worry about as long as you're using 24bit or more and you're not hearing any hiss (which is when you're actually hearing the noise floor of the DAC).
 

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
The better perceived clarity or dynamics could be caused by clipping, this is how loudness war mastering works. When people get used to hear such kinds of sounds, the non-clipping version could indeed sound worse to some people due to subjective preference.

A lot of studio interfaces can deliver more than 4Vrms without clipping so it is not necessary to limit yourself to the MQA models. If you already own more than one MQA DACs, I don't see why you cannot get a non-MQA one with higher non-clipping output for non-MQA content, unless you almost listen to MQA exclusively.

About that: "The better perceived clarity or dynamics could be caused by clipping." I must admit that this is the first time I have heard of such a thing. But the fact is that by gradually turning down the volume on the DAC I find a point where the clipping stops. Of course, the limit is not the same for all songs, but you can find a value that is ok for almost all of them. This would mean that the music is listened to at the limit of clipping. Of course, the closer the value is to -0dB, clarity and dynamics are better. But believe me the sound is incomparably better than with a filter or any other manipulation. Some albums, and there are quite a lot, can be listened to at -0dB, an incredible sense of clarity and power of voice, piano, saxophone, guitar..... If it is due to clipping or pre-clipping, thank them for that. Of course, everything depends primarily on the quality of the recording.

I love detailed and analytical sound, fast clear bass. That should be taken into account, because some people might not like it. Let's say the sound is similar to open studio headphones, as opposed to the sound of closed headphones with full bass and not so clear mid and upper spectrum.

By the way, these are the complete specifications of the SMSL D400ES

Input: USB/Optical/Coaxial/Bluetooth/IIS/AES
Output: RCA/XLR
Line output amplitude:
XLR 5.2Vrms
RCA 2.5Vrms
THD+N: 0.00005%(-125dB)(UN-WTD)
Dynamic range: 133dB(XLR),128dB(RCA)
SNR: 133dB(XLR),127dB(RCA)
USB transmission: USB 2.0 Asynchronization
USB compatibility:
Windows 7/8/8.1/10(Needs Driver)
Mac OSX10.6 or Later, Linux(Driverless)
MQA decoder: USB/Optical/Coaxial/AES
MQA-CD decoder: Optical/Coaxial/AES
Sampling rate & Bit Width:
USB PCM 44.1~768kHz(32bit)
DSD 2.8224~22.5792MHz(1bit)
Optical/Coaxial 44.1~192kHz(24bit)
DoP DoP256(USB),DoP64(Optical/Coaxial)
Bluetooth 5.0
LDAC: 24bit/96kHz (990kbps/660kbps/330kbps)
aptX-HD: 24bit/48kHz_576kbps
aptX: 16bit/44.1kHz_352kbps
SBC: 16bit/44.1kHz_328kbps
AAC: 16bit/44.1kHz_320kbps
Power Consumption: < 10W
Standby power: <0.5W
Size: 240x188x43mm (WxHxD)
Weight: 1.8kg/3.97Ib
 

waveuk

Member
Joined
Dec 22, 2022
Messages
17
Likes
3
About that: "The better perceived clarity or dynamics could be caused by clipping." I must admit that this is the first time I have heard of such a thing. But the fact is that by gradually turning down the volume on the DAC I find a point where the clipping stops. Of course, the limit is not the same for all songs, but you can find a value that is ok for almost all of them. This would mean that the music is listened to at the limit of clipping. Of course, the closer the value is to -0dB, clarity and dynamics are better. But believe me the sound is incomparably better than with a filter or any other manipulation. Some albums, and there are quite a lot, can be listened to at -0dB, an incredible sense of clarity and power of voice, piano, saxophone, guitar..... If it is due to clipping or pre-clipping, thank them for that. Of course, everything depends primarily on the quality of the recording.
Agree that no filter is how it sounds best for me.

However @amirm says that based on measurements, this is not something we would like to use as it eats too much into the audible spectrum.

Wondering if this is a case where perceived sound and objective measurements not fully align.
 
Last edited:

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
Agree that no filter is how it sounds best for me.

However @amirm says that based on measurements, this is not something we would like to use as it eats too much into the audible spectrum.

Wondering is this a cases where perceived sound and objective measurements not fully aligns.

Looking at the diagram Amir is right, but it's still better than bad filters.

Of course, we are still waiting for Amir's answer whether the measurements of 1 kHz tone at full volume using USB connection and XLR output were performed with filter On or OFF. That will clear things up a lot.
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
by gradually turning down the volume on the DAC I find a point where the clipping stops.
In this context it means the clipping stops being audible according to your ears, this is fine for personal listening, but it is not the same as the clipping stops being measurable. When quoting specs, this is important.

XLR 5.2Vrms
Assume this is real, and assume your DAC works in the same way as Amir's, then it means Amir was showing the SINAD value with the built-in filter enabled because Amir's result is only 4.2Vrms. Which means Amir was not intentionally set the filter to OFF in order to get a better SINAD value, but set the filter to ON when measuring other stuffs. This would be a very dishonest practice.

Turning off the filter without external upsampling means incorrect frequency response and a lot of distortion beyond 22.05kHz in 44.1kHz measurements, the audio analyzer will be able to detect these deficiencies in other measurements, those measurements require proper filtering to get good results, so the whole thing makes no sense.

The difference between 4.2Vrms and 5.2Vrms is only 1.855dB, but in the case of non-flat response this could be frequency dependent. You can confirm it by using this calculator:
52.png
42.png


Sometimes people plot graphs for readability / clarity and the displayed differences are not necessarily absolute, for example offsetting the frequency response plots so that they don't overlap, it doesn't mean some filters are louder or quieter than the others.
index.php
 
Last edited:

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
The only real thing that could happen when you use 64bit and reduce volume digitally is that you will lower the SNR (Signal to Noise Ratio) - this just means that if you lower it too much then you might start hearing hiss when music is not playing (very quiet or silent sections of the track). If you don't hear hiss then you're fine, you're not losing anything. 64bit or 32bit or 16bit, you still lose SNR when you use digital volume control. But using 32bit & 64bit (& 24bit if you don't reduce volume as much) will allow you to retain all those "volume steppings" that are included within your 16bit recorded music - which I think is described as quantisation error, so using (24bit)/32bit/64bit then you're protecting yourself from quantisation error and retaining all the "volume steppings" within your 16bit recorded music when using digital volume control. That's as much detail as I understand it. Basically, digital volume control is fine & nothing to worry about as long as you're using 24bit or more and you're not hearing any hiss (which is when you're actually hearing the noise floor of the DAC).

I didn't know that, thanks for the info.
 

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
In this context it means the clipping stops being audible according to your ears, this is fine for personal listening, but it is not the same as the clipping stops being measurable. When quoting specs, this is important.


Assume this is real, and assume your DAC works in the same way as Amir's, then it means Amir was showing the SINAD value with the built-in filter enabled because Amir's result is only 4.2Vrms. Which means Amir was not intentionally set the filter to OFF in order to get a better SINAD value, but set the filter to ON when measuring other stuffs. This would be a very dishonest practice.

Turning off the filter without external upsampling means incorrect frequency response and a lot of distortion beyond 22.05kHz in 44.1kHz measurements, the audio analyzer will be able to detect these deficiencies in other measurements, those measurements require proper filtering to get good results, so the whole thing makes no sense.

The difference between 4.2Vrms and 5.2Vrms is only 1.855dB, but in the case of non-flat response this could be frequency dependent. You can confirm it by using this calculator:
View attachment 258218View attachment 258219

Sometimes people plot graphs for readability / clarity and the displayed differences are not necessarily absolute, for example offsetting the frequency response plots so that they don't overlap, it doesn't mean some filters are louder or quieter than the others.
index.php

When I say "clipping stops" I definitely mean stops being audible according to my ears, not the clipping stops being measurable. Obviously, the clipping starts much earlier than I can hear.

I see that you are coming back to the fact that Amir was not intentionally set the filter to OFF in order to get a better SINAD value, and that no one ever mentioned it. (it was already said for the first time that no one has any doubts about Amir's work, which is highly appreciated not only on this forum). If I have any doubts about technical questions regarding measurements, I will always ask him because I know I will get a 100% accurate and precise answer. I hope that's clear, and we won't go back to it again.

But let's get back to the audio topic. I would be very grateful if you or someone else could guess what SMSL wanted with the filter OFF variant that cuts a good part of the audible spectrum (not to mention clipping), and then reduce the sound intensity by 4.2dB, then implement the filters. It seems very strange to me. I asked SMSL directly, but I didn't get a precise answer, in fact no answer at all. They just said to turn ON the filters or turn down the sound on the DAC with filter OFF. And they thanked me for understanding.
 

piaseczek

Member
Joined
Sep 18, 2022
Messages
39
Likes
22
But let's get back to the audio topic. I would be very grateful if you or someone else could guess what SMSL wanted with the filter OFF variant that cuts a good part of the audible spectrum (not to mention clipping), and then reduce the sound intensity by 4.2dB, then implement the filters. It seems very strange to me. I asked SMSL directly, but I didn't get a precise answer, in fact no answer at all. They just said to turn ON the filters or turn down the sound on the DAC with filter OFF. And they thanked me for understanding.
Setting filter off is a new feature built into the es9039pro. DAC is not doing the first step of oversampling so the signal is not attenuated by ~4dB. There is nothing strange in it.

These feature is introduced for external upsampling.
 

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
Setting filter off is a new feature built into the es9039pro. DAC is not doing the first step of oversampling so the signal is not attenuated by ~4dB. There is nothing strange in it.

These feature is introduced for external upsampling.

Thanks a lot for this information. So, they left the option for external upsampling. Although the curve for the OFF filter could have been better.
 

Aleksandar RS

Active Member
Joined
Dec 11, 2022
Messages
106
Likes
27
Here's how D400ES gets to 5.2Vrms (from the SMSL site).

1674160054309.png


Of course, they also made measurements at -0dB, but such measurements are already done here, so there is no need to copy them from the SMSL site.
 

waveuk

Member
Joined
Dec 22, 2022
Messages
17
Likes
3
Here's how D400ES gets to 5.2Vrms (from the SMSL site).

View attachment 258355

Of course, they also made measurements at -0dB, but such measurements are already done here, so there is no need to copy them from the SMSL site.
But we still do not know if those are with filters ON or OFF.

I would assume OFF, but if in other DACs this is not possible then like for like, has the D400ES worst results?
 

piaseczek

Member
Joined
Sep 18, 2022
Messages
39
Likes
22
I would assume OFF, but if in other DACs this is not possible then like for like, has the D400ES worst results?

Your assumption is wrong, digital filter is a part of DAC and turning it off for measurements doesn't have any sense.
 
Top Bottom