SMSL DO400 Fully Balanced Audio Decoder & Headphone Amplifier

[nocomments]

This is useful thanks. I'm also hyper treble sensitive and get ear pain for a couple days after wearing some of my too bright headphones. Would apodizing also be best for this? Thanks.

In this case, it makes sense to try "Filter Off".

 

[David_M]

Jitter Measurements on SMSL DO400:
Stimulus: REW JTEST, 48kS/s
Source for Coax and Toslink: SMSL PO100-AK
ADC: Cosmos ADCiso, 48kS/s

The spectra are pretty much clean.
- USB shows some lines at -146dBFS and a bit wider skirt
- Toslink shows a slightly broadened skirt with DPLL = 5 (default); with DPLL = 1 or 15 it looks better
- Coax shows a slightly broadened skirt with DPLL = 1; with DPLL = 5 or 15 it looks better

Hey...thank you very much for these important measurements ... now would you mind doing a full suite of measurements, bo9th for the HPA and line outputs, similar to Amirs and others? I saw measurements a while back of high distortion numbers from the HP connectors themselves, while they measure great at the internal HPA chips outputs, but once that signal is transferred to the output headphone connector board, distortion rises drastically.

I know the measurements task is quite a task and a half to do, but you'd be doing us a great service! Thanks again!

 

[nanook]

This is useful thanks. I'm also hyper treble sensitive and get ear pain for a couple days after wearing some of my too bright headphones. Would apodizing also be best for this? Thanks.

In case you hear the difference between #3 (Menu: "Apodizing") and #4 (Menu: "Linear Fast") I'm tempted to suspect Longfellow78 has got ears like a bat ;-)

Both, #3 and #4 are linear filters (pre and post- ringing in the time domain) and both of them are steep roll-off. The main difference is that #3 (apodizing) is ca. 7dB down at 20kHz whereas #4 is flat up to ca. 21kHz (valid for fs = 44100 kS/s; for higher fs all cut-off frequencies are higher)).
--> I would recommend apodizing #3 in case you really are sensitive fairly high up in the spectrum.
(But: what we perceive as "too bright" usually is too much in the frequency range around 3 ... 8kHz, where #3 and #4 are just equal.)

"Filter off":
This setting is meant for upsampling upstream in the data path (usually in software). This settings seems to do some low-pass filtering as well (this roll-off starting already at ca. 10kHz can not be the analog post-filter - it would be visible in the other frequency- response curves as well ). You can try this anyway, but please attenuate the DAC to at least -7dB, otherwise you may experience distortion (the filter-off setting is ca. 7dB louder than the others).

 

[nanook]

Hey...thank you very much for these important measurements ... now would you mind doing a full suite of measurements, bo9th for the HPA and line outputs, similar to Amirs and others? I saw measurements a while back of high distortion numbers from the HP connectors themselves, while they measure great at the internal HPA chips outputs, but once that signal is transferred to the output headphone connector board, distortion rises drastically.

I know the measurements task is quite a task and a half to do, but you'd be doing us a great service! Thanks again!

Thank you for the appreciation. I will post a couple of more measurements, but a full set is beyond my capability, time- and equipment wise.
Especially doing THD or IMD measurements vs. level is tough. I'm using the REW and this type of plot is not supported asfaik - at least I didn't find information on how to do this.
If someone out there can recommend a freeware (or shareware) for "distortion measurements versus level" that runs under W10, I'd be really pleased.

 

[arrow0309]

In case you hear the difference between #3 (Menu: "Apodizing") and #4 (Menu: "Linear Fast") I'm tempted to suspect Longfellow78 has got ears like a bat ;-)

Both, #3 and #4 are linear filters (pre and post- ringing in the time domain) and both of them are steep roll-off. The main difference is that #3 (apodizing) is ca. 7dB down at 20kHz whereas #4 is flat up to ca. 21kHz (valid for fs = 44100 kS/s; for higher fs all cut-off frequencies are higher)).
--> I would recommend apodizing #3 in case you really are sensitive fairly high up in the spectrum.
(But: what we perceive as "too bright" usually is too much in the frequency range around 3 ... 8kHz, where #3 and #4 are just equal.)

"Filter off":
This setting is meant for upsampling upstream in the data path (usually in software). This settings seems to do some low-pass filtering as well (this roll-off starting already at ca. 10kHz can not be the analog post-filter - it would be visible in the other frequency- response curves as well ). You can try this anyway, but please attenuate the DAC to at least -7dB, otherwise you may experience distortion (the filter-off setting is ca. 7dB louder than the others).

And for the bright lovers what do you reccomend? Which filter?
This new flagship ESS seems to me a little darker than the older ES9038PRO (x2 on my ex K9 Pro ESS)

 

[nanook]

And for the bright lovers what do you reccomend? Which filter?
This new flagship ESS seems to me a little darker than the older ES9038PRO (x2 on my ex K9 Pro ESS)

I'm afraid that PCM filters won't help. None of them will boost some frequency region.

It's actually funny, a friend of mine always complained the ESS DACs are too bright - he owns one with the ES9038q2m.
Looking at measurements, it's hard to believe that a ES9038 based DAC sounds different compared to eg. a good AKM chip based DAC.
I have heard this complaint "ESS DACs sound too bright" a couple of times and I 'd be very curious if this impression would "survive" blind testing.
In case it does, we will have to think about a new stimulus/analysis technique that can reveal and quantify these audible differences.

Edit: Measurement wise the ESS Hyperstream IV generation (ES9039xxx) has quite some improvements in the SD- modulator compared to the ES9038xxx (Hyperstream II). I would however be impressed if this is audible.

Detailed THD(+N) vs. Output Level Measurements of ESS ES9039Q2M. Balanced output quality from Hidizs AP80 PRO-X DAP. And Tekton's self-inflicted PR disaster.

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

The only recommendation I can give is "get an equaliser"
Honestly, most music I'm listening to has anyway been processed during production - and a slight EQ is a very mild processing to my opinion. I think it is allowed to adjust the sound such that you enjoy the music more - really nothing wrong with this.
The only caveat is that I would need to have different settings for almost every album....

 

[arrow0309]

I'm afraid that PCM filters won't help. None of them will boost some frequency region.

It's actually funny, a friend of mine always complained the ESS DACs are too bright - he owns one with the ES9038q2m.
Looking at measurements, it's hard to believe that a ES9038 based DAC sounds different compared to eg. a good AKM chip based DAC.
I have heard this complaint "ESS DACs sound too bright" a couple of times and I 'd be very curious if this impression would "survive" blind testing.
In case it does, we will have to think about a new stimulus/analysis technique that can reveal and quantify these audible differences.

Edit: Measurement wise the ESS Hyperstream IV generation (ES9039xxx) has quite some improvements in the SD- modulator compared to the ES9038xxx (Hyperstream II). I would however be impressed if this is audible.

Detailed THD(+N) vs. Output Level Measurements of ESS ES9039Q2M. Balanced output quality from Hidizs AP80 PRO-X DAP. And Tekton's self-inflicted PR disaster.

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

The only recommendation I can give is "get an equaliser"
Honestly, most music I'm listening to has anyway been processed during production - and a slight EQ is a very mild processing to my opinion. I think it is allowed to adjust the sound such that you enjoy the music more - really nothing wrong with this.
The only caveat is that I would need to have different settings for almost every album....

I see, I'm using Equalizer Apo (Peace) occasionally with some of my gear, it's only that I'm getting a little lazy
Maybe I'll buy some new phones instead, I'd like to try the HarmonicDyne Zeus Elite (for pc gaming as well).
Great stuff but I don't understand much from all those graphics lol

 

[David_M]

Thank you for the appreciation. I will post a couple of more measurements, but a full set is beyond my capability, time- and equipment wise.
Especially doing THD or IMD measurements vs. level is tough. I'm using the REW and this type of plot is not supported asfaik - at least I didn't find information on how to do this.
If someone out there can recommend a freeware (or shareware) for "distortion measurements versus level" that runs under W10, I'd be really pleased.

I completely understand ... how about THD+N vs frequency at standard 2Vrms (SE) or 4Vrms (DE) levels or even at the standard single 1KHz tone using USB/Optical/Coaxial inputs? Any available distortion tests would be welcome on the headphone SE and DE connectors.

Also, is there some way to pin all your measurements to the first page of this sub-forum so that everyone sees and knows about them?
Thanks again!

 

[nanook]

I see, I'm using Equalizer Apo (Peace) occasionally with some of my gear, it's only that I'm getting a little lazy
Maybe I'll buy some new phones instead, I'd like to try the HarmonicDyne Zeus Elite (for pc gaming as well).
Great stuff but I don't understand much from all those graphics lol

I just came across this review:
https://www.soundphilereview.com/reviews/smsl-do400-review-complete-power-43119/

They do mention the sound presets. I never tried them. Here is an excerpt from the manual:

SOUND COLOR
This machine has a total of 9 tone modes, which are processed by DSP through DAC, which are:
Rich1, Rich2, Rich3
Tube1, Tube2, Tube3
Crystal1, Crystal2, Crystal3
This setting is using the DSP of DAC, to make some different sound color, all of these settings will not change the frequency response curve. When set to "Standard", this setting will be bypassed.

I suspect that they are using the distortion compensation that is included in the chip and just dial in some relatively large "correction" values which leads to harmonic distortion.

This will likely not have the same effect you were describing for the difference between the K9 PRO ESS (ES9038PRO) and the ES9039PRO that you perceived. The distortion of the ES9038xxx is way beyond audibility just as that of the ES9039xxx (proper implementation presumed).

Just give it a try

 

[nanook]

Sound Color Modes in SMSL DO400:
The ESS DACs do have a THD compensation built into the digital data path (see figure copied from ES9039q2m datasheet). This compensation includes 2nd and 3rd harmonic only. The higher ones usually are below ca. -140dBc anyway.
From what I understand it modifies the linearity (in the digital domain) and thus allows to get a somewhat bent transfer function (in the SD-modulator and the external analog circuitry) straight again. As such it should work properly also at reduced signal levels.

I guess SMSL just utilizes this functionality as their Sound-Color modes modify 2nd and 3rd harmonic as well (see table below). All it needs is to write different values into 2 registers.

Generator frequency was 1 kHz. The measurement was performed with REW, APU (notch 1kHz, gain 26dB) and ADCiso (E1DA) in mono mode

Sound-Color H2 H3
Standard -136dBc -135dBc
Rich1 -96dBc -113dBc
Rich2 -82dBc -90dBc
Rich3 -74dBc -81dBc
Tube1 -96dBc -135dBc
Tube2 -82dBc -135dBc
Tube3 -74dBc -135dBc
Crystal1 -136dBc -113dBc
Crystal2 -136dBc -90dBc
Crystal3 -136dBc -81dBc

• Rich enhances H2 and H3
• Tube enriches H2 only
• Crystal enriches H3 only

I have not listened to these sound color settings yet and I'm afraid I will only hear the -74dBc ones - at best. To me these values seem pretty small.
The adjustment range in the DAC chip likely is restricted to what SMSL does here - finally it's meant to reduce and not to generate distortion ;-)
EDIT: In the ASR thread about "pkharmonic" users dial in values around -80dB ... -100dB. Seems my gut feeling may be wrong ....

In case you want more distortion, more flexibility and more conveniance, @pkane has developed a tool: https://distortaudio.org/pkharmonic.html

 

[nanook]

I completely understand ... how about THD+N vs frequency at standard 2Vrms (SE) or 4Vrms (DE) levels or even at the standard single 1KHz tone using USB/Optical/Coaxial inputs? Any available distortion tests would be welcome on the headphone SE and DE connectors.

Also, is there some way to pin all your measurements to the first page of this sub-forum so that everyone sees and knows about them?
Thanks again!

I have just started to play with the "Stepped Sine" capabilities of REW. Many things seem to be doable with this tool, some results seemed unexpected to me, so I guess I'm in the process of learning. The Multitone-Loopback-Analyzer would be another tool I can use.

Stepped level: I do not have a scaler, so the measurement limit for small levels will not be ideal. The level stepped sine with 1kHz I can do with the APU notch in place to keep the distortion of the ADCiso out as good as possible.
Stepped frequency: The AP uses a hardware notch filter afaik, so my measurements (ADCiso without notch) will not show the performance that we are used to from the AP

I'm not very much familiar with the forum software. I have no idea how I can maintain and extend a collection of measurements.

 

[Longfellow78]

In case you hear the difference between #3 (Menu: "Apodizing") and #4 (Menu: "Linear Fast") I'm tempted to suspect Longfellow78 has got ears like a bat ;-)

Both, #3 and #4 are linear filters (pre and post- ringing in the time domain) and both of them are steep roll-off. The main difference is that #3 (apodizing) is ca. 7dB down at 20kHz whereas #4 is flat up to ca. 21kHz (valid for fs = 44100 kS/s; for higher fs all cut-off frequencies are higher)).
--> I would recommend apodizing #3 in case you really are sensitive fairly high up in the spectrum.
(But: what we perceive as "too bright" usually is too much in the frequency range around 3 ... 8kHz, where #3 and #4 are just equal.)

"Filter off":
This setting is meant for upsampling upstream in the data path (usually in software). This settings seems to do some low-pass filtering as well (this roll-off starting already at ca. 10kHz can not be the analog post-filter - it would be visible in the other frequency- response curves as well ). You can try this anyway, but please attenuate the DAC to at least -7dB, otherwise you may experience distortion (the filter-off setting is ca. 7dB louder than the others).

Hey. Through some painful trial and error I have found that my ear pain is specifically related to around 3.5 to 3.6khz. I measured the frequency curves of headphones that hurt (t50rp) and those that don't (shp9500). The t50rp has a big spike at around 3.4 to 3.65k, and gives me pain listening to anything upper mid/ treble rich after even 15 seconds. Like the intro to another part of me by Michael Jackson. The shp on the other hand has no peak there, but a much higher peak at just over 4k, meaning my sensitivity is specific to that. I went into auto eq and ramped up that band to the max, it sounded very sparkly, and my ears are on fire around 2 hrs later. So that's the culprit. Will the apodising or linear fast settings help with this? I am going to eq my cans down to harman for the suspect frequencies. Thanks.

 

[nanook]

That sounds like you might have some flavour of Hyperacusis.
Using different PCM filters will not make a difference - in this frequency region the all are ruler-flat.
EQing sounds like a good solution. In case you indeed have some form of Hyperacusis, it might be that you have to adjust the EQ depending on listening volume.

 

[nanook]

SMSL D0400 THD 1kHz, 10kHz and Multitone
Distortion spectrum is as clean as we do expect from a proper ES9039xxx implementation.

THD:
The 1kHz and 10kHz spectrum were measured with a calibration file loaded into REW. Those were generated in loopback using DO400, APU, ADCiso. I.e. the dBc scale reflects the final numbers (without the need to account for the notch depth).
REW: 1kHz/10kHz, 0dBFS, 96kS
DO400: LinSharp, XLR
APU: notch 1kHz/10kHz, Gain 26dB
ADCiso: 96kS, 4V5, mono

Multitone:
The Multitone spectrum was measured with the DO400 driving the ADCiso directly in mono mode using a Y-splitter.
REW: Multitone
DO400:_LinearFast, XLR, 48kS
ADCiso: 48kS, 3v5, mono

 

[Longfellow78]

SMSL D0400 THD 1kHz, 10kHz and Multitone
Distortion spectrum is as clean as we do expect from a proper ES9039xxx implementation.

THD:
The 1kHz and 10kHz spectrum were measured with a calibration file loaded into REW. Those were generated in loopback using DO400, APU, ADCiso. I.e. the dBc scale reflects the final numbers (without the need to account for the notch depth).
REW: 1kHz/10kHz, 0dBFS, 96kS
DO400: LinSharp, XLR
APU: notch 1kHz/10kHz, Gain 26dB
ADCiso: 96kS, 4V5, mono

Multitone:
The Multitone spectrum was measured with the DO400 driving the ADCiso directly in mono mode using a Y-splitter.
REW: Multitone
DO400:_LinearFast, XLR, 48kS
ADCiso: 48kS, 3v5, mono

Thanks. Yeah sounds like "Noxacusis" for that frequency. It's funny as I have two sons and sometimes they scream really loudly and in a high pitch and I would go crazy because it hurt my ears so much and hurt for like an hour afterwards. My wife always said I was making it up and being a drama queen. Now I know this is around the maximum frequency a child can scream at. So it's the same thing!

​

Thanks for posting the graphs.. I can't really claim to understand what I seeing here though! I will try out the apodising one or linear fast as you suggest and see if i can tell the difference.
I assume we haven't heard anything about a new firmware from SMSL?

 

[arrow0309]

I just came across this review:
https://www.soundphilereview.com/reviews/smsl-do400-review-complete-power-43119/

They do mention the sound presets. I never tried them. Here is an excerpt from the manual:

SOUND COLOR
This machine has a total of 9 tone modes, which are processed by DSP through DAC, which are:
Rich1, Rich2, Rich3
Tube1, Tube2, Tube3
Crystal1, Crystal2, Crystal3
This setting is using the DSP of DAC, to make some different sound color, all of these settings will not change the frequency response curve. When set to "Standard", this setting will be bypassed.

I suspect that they are using the distortion compensation that is included in the chip and just dial in some relatively large "correction" values which leads to harmonic distortion.

This will likely not have the same effect you were describing for the difference between the K9 PRO ESS (ES9038PRO) and the ES9039PRO that you perceived. The distortion of the ES9038xxx is way beyond audibility just as that of the ES9039xxx (proper implementation presumed).

Just give it a try

I've just got the Zeus Elite headphones (for a few days) and WOW
Don't need any more colors lol
The dac / amp is just fine and the sound is (slightly V shaped) just amazing.
Both music and online gaming.
Maybe I'm gonna sell the Anandas v3, I really liked the older Ananda more.
If I ever come back to Hifiman will be with a pristine condition older model Arya.

 

[arrow0309]

Can anyone tell me what does the green dot MQA mean?
Because sometimes I get the blue dot (on Tidal).

 

[arrow0309]

I need to make another post for another issue (and it ain't new).
As soon as a 96K non MQA song is next on my Tidal (also anywhere else) or higher (even a 48K) I'll only get a small noise like a pop and then no sound.
Happening all the time I don't shut the dac down completely from the rear power switch, I mean the standby is so convenient, why SMSL?
Same exact issue I used to have with my ex DL300, there it was even worse due to missing rear on/off button so I had to just unplug the device.
That's soooo strange and Topping is a big step ahead (even with the automatic standby on the pc shut down or any other input).

 

[Longfellow78]

That sounds like you might have some flavour of Hyperacusis.
Using different PCM filters will not make a difference - in this frequency region the all are ruler-flat.
EQing sounds like a good solution. In case you indeed have some form of Hyperacusis, it might be that you have to adjust the EQ depending on listening volume.

Yeah so I don't have hyperacuisis I don't think, as it doesn't relate to regular volume sounds only headphones. the mystery deepens however. I just spent days and days modding my T50RPs, even removing rear driver paper and replacing it with different types of breathable tape. Finally killed the 3.5k high so its the same level as my shp9500. And..... ears still hurt! WTF. I'm wondering if it's something crazy like to do with planar magnetics - seems impossible. Thinking back to all the headphones I've had. EU6000, HyperX, SHP9500, T50RP, T60 Argons, the only ones that give me pain are the Fostex. I'm done with the bastards, going to sell them.

 

[nanook]

Yeah so I don't have hyperacuisis I don't think, as it doesn't relate to regular volume sounds only headphones. the mystery deepens however. I just spent days and days modding my T50RPs, even removing rear driver paper and replacing it with different types of breathable tape. Finally killed the 3.5k high so its the same level as my shp9500. And..... ears still hurt! WTF. I'm wondering if it's something crazy like to do with planar magnetics - seems impossible. Thinking back to all the headphones I've had. EU6000, HyperX, SHP9500, T50RP, T60 Argons, the only ones that give me pain are the Fostex. I'm done with the bastards, going to sell them.

Well, keep in mind that listening with headphones usually is at higher levels - as far as I remember the listening level with headphones is typically ca. 10dB louder compared to speakers (I never tried to understand why, though).