SMSL SU-9 Balanced DAC Review

[voodooless]

I hate to bring this up once more, but what's up with that attenuation again? None of the ESS DAC's have this kind of lowish attenuation. We've seen it now multiple times that the measurement do not comply with DAC spec. It's also ruler flat for all filters, while stop band specs and spectrums are vastly different for the various filter options. Something must be up with that! @amirm Is this now a measurement artifact? If not, how can this be? I guess they could have loaded custom filters.. but why the effort, since the default ones clearly have better performance (if measurement is correct).

 

[Veri]

I hate to bring this up once more, but what's up with that attenuation again? None of the ESS DAC's have this kind of lowish attenuation. We've seen it now multiple times that the measurement do not comply with DAC spec. It's also ruler flat for all filters, while stop band specs and spectrums are vastly different for the various filter options. Something must be up with that! @amirm Is this now a measurement artifact? If not, how can this be? I guess they could have loaded custom filters.. but why the effort, since the default ones clearly have better performance (if measurement is correct).

The chip filters are configurable. They are still the default loaded filters, but they also have their own settings. So once more, it is entirely up to implementation. In any case, it looks good? Proper Nyquist fs/2 filters are available, but a little lacking in attenuation yes. Looks fine to me...

 

[KxDx]

As an Englishmen's signature on this site indicates "Too many DACs> ".
So many dacs have not only about same measurements today; but also same features.
So how to choose, only by it's style or ... ?

Style. Features, such as a headphone amp or bluetooth.

I bought a TEAC UD-505 because I got a killer deal and I love the way it looks. The voice in my head keeps telling me to sell it for a $400 profit and buy something cheaper that's 100% as functional, but it's a beauty. eh, who knows?

 

[vkvedam]

So many choices, we should be doing subjective assessments going forward I suppose after consulting the audiologist

 

[AudioSceptic]

This is a review and detailed measurements of the brand new SMSL SU-9 Balanced USB DAC. It was kindly sent to me by one of their distributors, Shenzhenaudio. The SU-9 will be released September 30th and will cost US $439.99.

The SU-9 has a simpler but in my opinion more elegant look than previous SMSL DACs:

View attachment 84224

The display is now larger and courtesy of its fine resolution, nice to look at. The rotary control lets you navigate the menus by pressing and rotating. Alas, I could not figure out how to get out of the menus and had to wait for it to time out. Perhaps there is a way but not having read the manual, I did not know how. There is a remote of course which would make this simpler.

Back panel is what you expect in this class DAC plus addition of Bluetooth:

View attachment 84225

Power supply is included of course eliminating the external bricks that come with lower cost DACs. And of course balanced XLR output which I highly encourage for computer connected DACs.

DAC Audio Measurements
As usual, we start with our dashboard view:

View attachment 84226

Wow! Another DAC nearly maxing out the distortion+noise meter in my analyzer! The SINAD of 120 dB puts the SU-9 way up there in rankings:
View attachment 84227

Zooming in:

View attachment 84228

Note that the SU-9 is capable of output above 4 volts which is nice. To show the performance at other output levels, I ran a sweep:

View attachment 84229

Since the SINAD measurement is noise limited (distortion alone is at 130 dB SINAD), having the output higher helps improve it as you see it in the above graph.

IMD distortion versus level shows excellent performance:

View attachment 84230

Our old friend the "ESS IMD Hump" is still there but 90% suppressed.

Linearity which is a measure of accuracy is perfect:

View attachment 84231

Jitter over USB is non-existent:
View attachment 84232

There is a digital PLL that helps reduce jitter but since there is none with USB input, it makes no difference. It however helps with Toslink jitter:
View attachment 84233

There may be a latency penalty though so if you are using Toslink for TV sound input, you may need to play with that. Fortunately the jitter that was there with default setting was not an audible concern anyway (at -120 dB and heavily masked by the signal itself).

Lots of filters are provided for any taste (imagined or otherwise):
View attachment 84235

I like that there is a proper brickwall filter. I used that for the following test:

View attachment 84236

There is some frequency dependency here. I ran an FFT (not shown) and it showed that to be "imaging" components that were changing way up in frequency so not an audible issue.

Finally, multitone test showed how low the distortion is in SU-9 DAC:

View attachment 84237

Conclusions
Looks like we started a trend to achieve perfection in desktop DACs and the SMSL SU-9 is yet another example of that. We have another great "instrument grade" DAC on our hands providing absolute transparency to the source beyond any threshold of human hearing. If you are hearing issues I suggest you either have your hearing examined, or learn to do a proper controlled test. Either way, what you hear is what is on your recording. Guaranteed!

It is my pleasure to highly recommend the SMSL SU-9. Great job.

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

I explained last night that I poked my palm last night trying to take the core out of Asian Pears in order to dry them. I swore I would not do any more but then I tasted the results this morning. Ah, they were fantastic! Sweet, full of pear flavor and all fit in a nice jar taking a fraction of space to store. So I dragged the bucket of pears out of the garage and prepared another batch to dry:

View attachment 84239

Despite filling 8 stacks I hardly made a dent in all the pears we have. And we have yet to pick the pears from two of trees!

Dehydrating uses electricity and while our rates are some of the lowest in the nation/world, it is not free. So appreciate some good sized donations to pay for it using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

I see the filters are the same as for the Pro-Ject Pre Box S2 Digital, except the Pro-Ject has one extra option <https://www.audiosciencereview.com/...ct-pre-box-s2-dac-headphone-amp-review.10085/>. Although the SMSL measures much better in other ways, the Pro-Ject has greater (better) filter attenuation. Why would this be?

 

[Richdw]

This looks to be a great device, and follows on nicely from the similar Evo Sabre reviewed last week. They're similar money, although I've not seen UK/EU pricing yet. The Evo Sabre has more input options, potentially making it more functional as a system hub (e.g. if you want to plug in an aux cable etc), and a streaming/room-correction RPi included. However, SINAD is superior and the ESS hump is relatively lower on the su-9, yetcgiven the dearth of ABX listening tests it is still not apparent whether that difference is audible.

Tech query:
@amirm what bluetooth codec/codecs are used? The codec type, and/or ability to select different codec types is quite important when BT connectivity is a necessary factor. Could codec options be included in future reviews, where it is easy to determine?

Many thanks, Richard

 

[Jimbob54]

So why do we need to pay for M400? (well, nobody needs to, but what does the $400 buy you?)

 

[JohnYang1997]

So why do we need to pay for M400? (well, nobody needs to, but what does the $400 buy you?)

Multitone is actually better here. Low frequency distortion on M400 is quite high comparing to 1khz.

 

[w1000i]

It has DSD filter and sound color. Is it in ESS or programed by SMSL

 

[Jimbob54]

Multitone is actually better here. Low frequency distortion on M400 is quite high comparing to 1khz.

Last thing I need is another DAC of course....just not sure on the sense of the product lineup and pricing.

 

[renaudrenaud]

So why do we need to pay for M400? (well, nobody needs to, but what does the $400 buy you?)

M400 is more organic.

 

[Veri]

M400 is more organic.

Orgasmic? I'd pay for that.

 

[voodooless]

The chip filters are configurable. They are still the default loaded filters, but they also have their own settings. So once more, it is entirely up to implementation. In any case, it looks good? Proper Nyquist fs/2 filters are available, but a little lacking in attenuation yes. Looks fine to me...

I don't buy it. The filters are not configurable, you can just choose one of 7, or use your own FIR coefficients. Like I said, why would you go though that trouble of making 7 new ones if you already have 7 to choose from? It would make absolutely no sense.

 

[Jimbob54]

M400 is more organic.

No pesticides clogging up the filters ?

 

[mertvp]

Amirm:

Thanks for your great work. This site is a wonderful counter to the snake oil shills at Stereophile and Absolute Sound.

I'd like to suggest that you include an SMPTE versus output level plot in you power amplifier reviews as you do in DAC reviews.
The SMPTE plot is vastly more revelatory of design quality and listenability than a 1khz distortion plot.

P.S. I've received no confirmation that my donation to your cause was disposed as I intended.

 

[Veri]

I don't buy it. The filters are not configurable, you can just choose one of 7, or use your own FIR coefficients. Like I said, why would you go though that trouble of making 7 new ones if you already have 7 to choose from? It would make absolutely no sense.

ESS datasheets tend to be confidential, but I don't really know what there is to deny. They're most certainly configurable, just like THD compensation. Wouldn't it mean SMSL is just using their own coefficients? I'm not saying it makes sense but it's one possible explanation.

Cirrus Logic also allows additional configuration of their filter presets. (single speed, double speed, quad speed modes) ... not sure about AKM.

 

[bennetng]

ESS datasheets tend to be confidential, but I don't really know what there is to deny. They're most certainly configurable, just like THD compensation. Wouldn't it mean SMSL is just using their own coefficients? I'm not saying it makes sense but it's one possible explanation.

Cirrus Logic also allows additional configuration of their filter presets. (single speed, double speed, quad speed modes) ... not sure about AKM.

 

[Veri]

View attachment 84261

If all manufacturers are using the exact same 7 pre-programmed digital filters it would be weird that they show differing attenuation strength each time I feel like we're missing something.

 

[sonci99]

It would be great to see some tear down of these dacs. Apart from the measurements I've always considered SMSL built quality not at Topping level..

 

[mertvp]

Amirm:

Thanks for your great work. This site is a wonderful counter to the snake oil shills at Stereophile and Absolute Sound.

I'd like to suggest that you include an SMPTE IM vs output level plot in your power amplifier reviews as you do in the DAC
reviews. The IM plot is vastly more revelatory of design quality and listenability than a 1khz distortion plot.