• 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 M400 Balanced USB MQA DAC Review

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
This is a review and detailed measurements of the SMSL M400 balanced and unbalanced DAC. It was kindly sent to me by the company. The M400 costs US $810 from prices I see online.

The M400 design has been upgraded with a larger, more clear display and a Plexiglas upper cover:

SMSL M400 USB DAC XLR Balanced Audio Review.jpg
There is the usual remote control you see across a wide range of products from China. I like it normally but in this instance, it did not have proper key debounce and as such, it would repeat the key someones making it harder to make selections.

The back panel is as you expect with the exception of missing AES/EBU balanced digital XLR input:

SMSL M400 USB DAC XLR Balanced Back Panel Inputs Outputs XLR RCA Audio Review.jpg

There is good bit of heft to the unit, giving a feeling of quality.

As you see there is also bluetooth input which I did not test.

MQA decoding is included although I did not test that.

DAC Audio Measurements
As usual we start with our dashboard of 1 kHz and to be fair, with devices that output more than 4 volts balanced, I adjust the output down to get 4 volts:

SMSL M400 USB DAC XLR Balanced Audio Measurements.png


Wow, this is excellent performance. Noise floor is about to fall off the chart on FFT. Distortion is less than -135 dB and as low as -140 dB in one channel. With best case hearing threshold being around -115 dB, we are way, way past transparency here.

SINAD which is a score that sums distortion and noise is much lower than the distortion, limited by the noise output of the M400 and my analyzer at 121 dB. Still, we have a tie for top spot of all DACs ever tested regardless of price or configuration:

Best balanced USB DAC Review Measurement 2020.png


Letting the DAC go to its maximum output improves the SNR of both the analyzer and DAC allowing even better SINAD:

SMSL M400 USB DAC XLR Balanced Full Output Audio Measurements.png


We can see it also as we step through the levels:

SMSL M400 USB DAC XLR Balanced THD+N SINAD vs Level Audio Measurements.png


RCA output is almost as good:

SMSL M400 USB DAC XLR Unbalanced RCA Audio Measurements.png


Signal to noise ratio as stated, is more or less at the limit of what we can measure:

SMSL M400 USB DAC XLR Balanced Dynamic Range Audio Measurements.png


Linearity is absolutely perfect even if I reduce the output to 4 volts:

SMSL M400 USB DAC XLR Balanced Linearity Audio Measurements.png


Multitone follows suite with also superb level of intermodulation distortion:

SMSL M400 USB DAC XLR Balanced Multitone Audio Measurements.png


IMD is also excellent as well:
SMSL M400 USB DAC XLR Balanced IMD Audio Measurements.png


The very low noise though allows us to see a tiny step up in intermodulation distortion at levels higher than -40 dB.

Jitter over USB is excellent:

SMSL M400 USB DAC XLR Balanced Jitter Audio Measurements.png


Toslink is not in the same league though:

SMSL M400 USB DAC XLR Balanced Toslink and Coax Jitter Audio Measurements.png


Company says this is due to the connector and better ones are no longer in production. Coax input is better with far less jitter although still a step behind USB. Either way, don't let the graphs mislead you into thinking there is an audible problem here. Even with Toslink it should not reduce audible performance.

Using 90 kHz bandwidth so we can capture harmonics of 20 kHz tone, we have very good performance but not perfect:

SMSL M400 USB DAC XLR Balanced THD+N vs Frequency Audio Measurements.png


Reason for that is that the reconstruction filter while better than many DACs, still allows out of band signals bleed through:

SMSL M400 USB DAC XLR Balanced Filter Frequency Response Audio Measurements.png


SMSL M400 USB DAC XLR Balanced 10 kHz FFT Audio Measurements.png


Nothing audible to worry about but my plea continues for DAC chip companies to provide better attenuation of out of band signals.

Output Impedance
I did not get a chance to measure output impedance but from data I have seen elsewhere, it rises to well over 1000 ohm. This is about 5 to 10 higher than it is normally. An output buffer stage has been eliminated to improve measured performance with the cost being that output impedance goes up. When you think of pairing this DAC with headphone and power amplifiers, make sure their *input* impedance is at least 10 times higher or 10 kHz and above. You should be OK with many products but with others, you may see a voltage drop. Seeing how M400 has more than 4 volts output anyway, maybe this is not a concern in practice.

Conclusions
We have another "instrument grade" DAC here from SMSL in the form of M400. Performance is beyond question, landing at the top of the class. Price has gone up with it as well but there is no free lunch. :)

On a slightly down note, it seems the race for the top place in our ranking is resulting in some design decisions such as removal of output buffer and increase of output impedance. I am not too happy about this but haven't decided what to do about it.

Anyway, I am happy to recommend the SMSL M400.

-----------
As always, questions, comments, recommendations, etc. are welcome.

Took the panthers out to the park for them to stretch their legs. A park ranger was there and gave me a ticket because they were not social distancing! Now I have that bill to play. As such, I appreciate you donating using : https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
 

ClosDeLaRoche

Active Member
Joined
Jun 21, 2019
Messages
136
Likes
131
"An output buffer stage has been eliminated to improve measured performance with the cost being that output impedance goes up."

Uhh what? This sounds concerning. Don't most DACs have a similar output impedance? This one is 5 to 10 times higher to hit better numbers elsewhere?
 

rebbiputzmaker

Major Contributor
Joined
Jan 28, 2018
Messages
1,099
Likes
463
On a slightly down note, it seems the race for the top place in our ranking is resulting in some design decisions such as removal of output buffer and increase of output impedance.

I am not too happy about this but haven't decided what to do about it.

So manufactures will discard best practice in design in order to obtain your reviewer blessing.

Time to run for political office.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
"An output buffer stage has been eliminated to improve measured performance with the cost being that output impedance goes up."

Uhh what? This sounds concerning. Don't most DACs have a similar output impedance? This one is 5 to 10 times higher to hit better numbers elsewhere?
Correct. A buffer is an amplification stage which doesn't increase the level but reduces impedance in this case. This stage has been eliminated resulting in higher output impedance. It now requires to pay attention to what you hook it up to.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
So manufactures will discard best practice in design in order to obtain your reviewer blessing.

Time to run for political office.
Not catering to me personally but rather, trying to produce absolute lowest noise and distortion.
 

ReaderZ

Addicted to Fun and Learning
Joined
Apr 14, 2020
Messages
618
Likes
414
Output Impedance
I did not get a chance to measure output impedance but from data I have seen elsewhere, it rises to well over 1000 ohm. This is about 5 to 10 higher than it is normally. An output buffer stage has been eliminated to improve measured performance with the cost being that output impedance goes up. When you think of pairing this DAC with headphone and power amplifiers, make sure their *input* impedance is at least 10 times higher or 10 kHz and above. You should be OK with many products but with others, you may see a voltage drop. Seeing how M400 has more than 4 volts output anyway, maybe this is not a concern in practice.

This is very significant, this marked for the first time, ASR's existence has made a product worse and not better. This of course cannot be blamed on ASR in any way, but should be cautioned and maybe consider a rating system that focus more on points like this. Considering 10k input is standard on many studio monitors and M400 can go well over 1000 ohm, this means many will not achieve the 10x higher number,
 

AudioJester

Addicted to Fun and Learning
Forum Donor
Joined
Jan 7, 2020
Messages
912
Likes
1,211
So if you measured an amp with 10k input impedance being fed by this dac, what would you expect to see due to impedance "mismatch"? Would it lead to power reduction?
 

Cahudson42

Major Contributor
Joined
Sep 21, 2019
Messages
1,083
Likes
1,556
Personally, no way am I going to pay $800 (or more) for a DAC. Perfectly happy with my $100 LG V20, and get streaming to boot.

We need another DAC just about as much as we need yet another streaming service (HBO Max) - which is NOT..

Fewer DACs taking up Amir's time would be my preference.
 
Last edited:

Seraphyx

New Member
Joined
Jun 1, 2020
Messages
3
Likes
3
Correct. A buffer is an amplification stage which doesn't increase the level but reduces impedance in this case. This stage has been eliminated resulting in higher output impedance. It now requires to pay attention to what you hook it up to.
Totally noob question but is this equivalent of turning off the preamp mode to be DAC only on the Topping D90?
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
So if you measured an amp with 10k input impedance being fed by this dac, what would you expect to see due to impedance "mismatch"? Would it lead to power reduction?
10k input is mostly fine. Even with lower input impedance you mostly only see a voltage drop. As mentioned, the already 5.3V is higher then normal hence it doesn't really matter if you get a voltage drop like that.
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
I'm not convinced there is a missing output buffer stage. This picture of the PCB shows the output. There is a 100Ω resistor in series with what looks like 510Ω* on the outputs. The RCA outputs are clearly driven by OPA1611 op amps, and the DAC is buffered by dual versions of the same. Both RCA and balanced have the same componentry on the output. (It looks like maybe some sort of anti-thump).

* The colours are awful, so it is hard to be sure, 510 is a standard E24 value, so seems reasonable. The resistor is a 1% part, but 510 isn't an E96 value - but you often see 1% in E24 ranges, so I assume this is the value.
 

Attachments

  • SMSL-400_close_up.png
    SMSL-400_close_up.png
    2.3 MB · Views: 4,696

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
I'm not convinced there is a missing output buffer stage. This picture of the PCB shows the output. There is a 100Ω resistor in series with what looks like 510Ω* on the outputs. The RCA outputs are clearly driven by OPA1611 op amps, and the DAC is buffered by dual versions of the same. Both RCA and balanced have the same componentry on the output. (It looks like maybe some sort of anti-thump).

* The colours are awful, so it is hard to be sure, 510 is a standard E24 value, so seems reasonable. The resistor is a 1% part, but 510 isn't an E96 value - but you often see 1% in E24 ranges, so I assume this is the value.
The XLR outputs aren't. And the output impedance should be the sum of two branches.
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
The XLR outputs aren't. And the output impedance should be the sum of two branches.
True, but we don't have an exact value for the output impedances. The sum would get us 1.2kΩ, which is about what is suggested. The 1611s on the unbalanced are clearly just creating an unbalanced from the balanced signal. But the 1612s should be otherwise the same drive capability.
It seems odd having so much resistance on the output. I guess they are current limiting the output. But odd none-the-less.
All I'm looking at is the specific suggestion that there is an explicit and unusual lack of output buffer that is characterised by the high output impedance. So far as I can see that output impedance is completely explained by the choice of passive components on the output. Whether this is a good idea is another matter.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
True, but we don't have an exact value for the output impedances. The sum would get us 1.2kΩ, which is about what is suggested. The 1611s on the unbalanced are clearly just creating an unbalanced from the balanced signal. But the 1612s should be otherwise the same drive capability.
It seems odd having so much resistance on the output. I guess they are current limiting the output. But odd none-the-less.
All I'm looking at is the specific suggestion that there is an explicit and unusual lack of output buffer that is characterised by the high output impedance. So far as I can see that output impedance is completely explained by the choice of passive components on the output. Whether this is a good idea is another matter.
This design of passive LPF is in the datasheet of ak4499. But I would argue it's for paralleling the outputs. The noise performance of this particular implementation of 4499 is not too great (1.6uV)that a opa1612 buffer (0.2uV) would cause any degradation. Not even mentioning half of 4499 seems to be turned off (lacking half of the components for a complete 4499 implementation).
 
Top Bottom