SMSL D-6s Balanced DAC Review

[HalSF]

General information for all:

The Probable Corrected List of the ESS Filters

The reality for your D6S likely looks like this:

• FL1 (Incorrect): Labeled as "Minimum Phase", but is actually Minimum Phase Slow (ESS #8).
• FL2 (Correct): Linear Phase Apodizing (ESS #2).
• FL3 (Correct): Linear Phase Fast (ESS #3).
• FL4 (Correct): Linear Phase Fast Low Ripple (ESS #4).
• FL5 (Correct): Linear Phase Slow (ESS #5).
• FL6 (Correct): Minimum Phase Fast (ESS #6).
• FL7 (Correct): Minimum Phase Slow (ESS #7).

the original ESS #1 filter is missing from the selection.

The ES9039 chip technically offers 8 different filters (ESS #1 through ESS #8). Since the SMSL D6S menu only provides 7 slots (FL1 to FL7), one filter had to be omitted.

SMSL decided to assign ESS #8 to the FL1 slot. As a result, the original ESS #1 ("Minimum Phase Fast Roll-off") is not available in the menu.

Does this matter?Not really, because:The filter FL6 (ESS #6) is also a "Minimum Phase Fast Roll-off". In terms of technical performance and sound characteristics, ESS #1 and ESS #6 are extremely similar. Both offer no pre-ringing, a sharp impulse response, and a steep roll-off.

If you are looking for the sound that the missing ESS #1 would have provided, simply use FL6. You aren't missing out on any specific sound signature; the labeling is just confusing.

Alrigthy?

Thanks for this update. This business with the filter settings seems sort of bumbling and comic because it never seemed to be correctly documented and yet also never seemed to matter at all.

But may I ask, just to keep the fun alive: what is your source for this probable, likely update info?

 

[umeng2002]

General information for all:

The Probable Corrected List of the ESS Filters

The reality for your D6S likely looks like this:

• FL1 (Incorrect): Labeled as "Minimum Phase", but is actually Minimum Phase Slow (ESS #8).
• FL2 (Correct): Linear Phase Apodizing (ESS #2).
• FL3 (Correct): Linear Phase Fast (ESS #3).
• FL4 (Correct): Linear Phase Fast Low Ripple (ESS #4).
• FL5 (Correct): Linear Phase Slow (ESS #5).
• FL6 (Correct): Minimum Phase Fast (ESS #6).
• FL7 (Correct): Minimum Phase Slow (ESS #7).

the original ESS #1 filter is missing from the selection.

The ES9039 chip technically offers 8 different filters (ESS #1 through ESS #8). Since the SMSL D6S menu only provides 7 slots (FL1 to FL7), one filter had to be omitted.

SMSL decided to assign ESS #8 to the FL1 slot. As a result, the original ESS #1 ("Minimum Phase Fast Roll-off") is not available in the menu.

Does this matter?Not really, because:The filter FL6 (ESS #6) is also a "Minimum Phase Fast Roll-off". In terms of technical performance and sound characteristics, ESS #1 and ESS #6 are extremely similar. Both offer no pre-ringing, a sharp impulse response, and a steep roll-off.

If you are looking for the sound that the missing ESS #1 would have provided, simply use FL6. You aren't missing out on any specific sound signature; the labeling is just confusing.

Alrigthy?

Yes, I was reading the manual, amirm's review, and the DAC chip's data sheet, and none of them seem to line up. Although your ESS numbering seems to be different than from what I found. Maybe I used the wrong data sheet. The SMSL online manual claims FL7 is a Minimum Phase Slow Roll Off Low Dispersion filter, but amirm's measurements indicate a fast roll off. So really ESS #7 (from the data sheet I have). My paper manual omits the low dispersion qualifier.

I'm enjoying this DAC quite a bit. Nice upgrade from my 16-year-old little Dot DAC_I

 

[umeng2002]

RIP to the color blind.

Online PDF manual v1.2:

Paper manual v1.0:

amirm review:

After looking at it a bit, amirm's review and the paper manual ver 1.0 filters line up, sans the filter word description from the paper manual. Ignore words. Trust your eyes and ears. Or just oversample.

I don't speak Chinese, nor does my computer; but I flashed my new unit with whatever was on SMSL's website (a somewhat more complex process than what you would think - at least the instructions were in English). I'm sure it came with the latest firmware, and I'm sure the website actually hosts the latest firmware (manual ver 1.2); but I did it anyway in case I got a really old stock unit.

Yes, based on the latest ver. 1.2 manual, I would pick FL6 as long as your unit has the most up to date firmware. FL3 is good, but not minimum phase.

 

[Kosmonaut]

Due to the lack of time long story short:

SMSL filters mislabeled on ES9039 DACs

Could you post this in English? Hello, I have been reading this topic for several days and it is extremely interesting. I have recently acquired a DO100Pro and unfortunately I am having trouble understanding everything about the bad labeling of the filters. I use it in XLR on a Nuprime Sta9 and...

www.audiosciencereview.com

Perhaps someone could explain what a digital reconstruction filter is and why, in relation to the SMSL D-6s, it's relatively unimportant which one you choose. Unless you have Vulcan gold ears.

Have fun!

 

[umeng2002]

Think of a hard frequency response filtering as a wall. You hit the wall, and get reflection back. So you need some sort of damping or attenuation so it reflects less than what is audible.

In more technical terms, a discrete sample has infinite bandwidth and therefore the "original" audio on the digital recording has information from those frequencies out to infinity. So anything above double the sampling rate needs to be filtered out to reproduce the original audio analog sound wave.

This is also why oversampling hides distortion caused by the different filtering techniques, other than time-domain issues, I believe. most of the filters distortion occurs close to the cut off. Over sampling drives it way beyond hearing range. Really, these filters only matter when playing 44.1 kHz material natively at 44.1 kHz... and you have really well trained ears. I would say, in the modern time, the impulse response is the biggest difference.

 

[RandomEar]

Think of a hard frequency response filtering as a wall. You hit the wall, and get reflection back. So you need some sort of damping or attenuation so it reflects less than what is audible.

Not sure this is the best example, because you filter precisely to avoid this effect. You would only see the aliasing if you skipped the filter.

In more technical terms, a discrete sample has infinite bandwidth and therefore the "original" audio on the digital recording has information from those frequencies out to infinity. So anything above double the sampling rate needs to be filtered out to reproduce the original audio analog sound wave.

* above half the sampling rate

The original audio should also be filtered already, but I agree that it will need infinite bandwidth if you try outputting it with the NOS / sample & hold method.

This is also why oversampling hides distortion caused by the different filtering techniques, other than time-domain issues, I believe. most of the filters distortion occurs close to the cut off. Over sampling drives it way beyond hearing range. Really, these filters only matter when playing 44.1 kHz material natively at 44.1 kHz... and you have really well trained ears. I would say, in the modern time, the impulse response is the biggest difference.

Filtering after oversampling is not about distortion, but about avoiding aliasing in the audible range and ultrasonic images. It's true that this matters less for - say - 96 kHz content, because the higher Fs pushes any aliasing far away from the audible range and into a territory where even most tweeters aren't bothered anymore. Though I would not skip oversampling for 48 or 96 kHz content - it's just better to keep it active.

The impulse response of filters always was their biggest difference, because the impulse response fully defines the filter

More about filtering here.

 

[umeng2002]

Yes, thanks for the more technical explanation. I always forget these details after awhile since I'm not formally trained in signal processing.

FYI, go to ShenzenAudio's website for the English version of the firmware updater program. SMSL only has the Chinese language version.

 

[Yviena]

I have received my D-6s in the meantime, yet I did not have time to listen to it extensively.

About the SMPS:
On one hand it's quite handy to have the power supply inside the unit (no bulky wall-wart that blocks several positions in the mains stripe), on the other hand this always raises issues with "noise" - and yes, this applies to SMPS as well as old fashion transformer based power supplies:

Transformers suffer from:
- stray magnetic field (especially these nice flat molded transformers need a lot of distance or magnetic screening inside the unit and schottky diodes in the rectifier and maybe even snubbering)
- leakage current that generates I*R-drop in the signal cables downstream and thus "hum" (ideally they should have a foil- shield between the primary and the secondary windings)

SMPS suffer from:
- relatively high leakage current with a lot of harmonics of the mains frequency (I*R-drop just like above)
- high frequency content that usually is beyond the audio spectrum (might get mixed down in case it finds its way into clocks)

You can get rid of the leakage current by connecting the audio- ground to PE (protective earth). This is done in the D-6s as PyramidElectric just mentioned above: RCA- shield and XLR pin1 are connected to PE.
This however gives rise to ground-loops (e.g. if the amplifier is grounded as well) and when magnetic fields are around (big transformers) this will very likely cause audible hum when using an asymmetric connection (RCA). With XLR you will still be fine because the induced voltage shows up common-mode and thus gets subtracted; this is why the pro-stuff uses differential signalling.

In my system the D-6s is the only grounded unit so I'm fine and the D-6s is dead quiet indeed.

My preference would always be an external power supply, ideally transformer based with shield:
- even if it's an SMPS I can replace it by a good transformer based one
- the SMPS is usually the first thing that breaks down (usually caps dry out due to ripple current -> temperature)

I like the solution that Topping has chosen on the L30II. This unit has a (bulky) transformer as wall-wart (such that the stray magnetic fields are away from the unit) and the unit is fed by AC voltage. The leakage current is small compared to an SMPS and the harmonic content is less audible. When the wiring of rectifier and caps is done properly inside the unit, there will not be much to complain about.
Even better, rectifier and first set of caps would be in the transformer (such that the current spikes that recharge the caps generate their magnetic field outside the unit), but you'd need a 3-wire connection then, which adds cost.

Kind of a old post but i just got the D6s and i'm using it with a schiit midgard amp (unsure if grounded) with 3m rca cables would xlr cables be better in this case?

 

[ICIETDIYEUR]

Hi.

Connections via XLR are always better than those via RCA for several reasons -> GOOGLE is your friend (or ask an IA)

 

[somebodyelse]

Connections via XLR are always better than those via RCA for several reasons -> GOOGLE is your friend (or ask an IA)

Well...they ought to be, but there are some examples of terrible engineering out there so it's not guaranteed.

 

[antcollinet]

Hi.

Connections via XLR are always better than those via RCA for several reasons -> GOOGLE is your friend (or ask an IA)

would xlr cables be better in this case?

XLR (or other balanced connections) typically provide higher voltage, and reject common mode noise.

If you don't have any audible noise, and you don't need the voltage, then unbalanced RCA is just as good.


Although with such long cables, it is quite likely for there to be a certain amount of common mode noise.

 

[Yviena]

Hmm I see may as well get some 3m xlr cables then.

 

[antcollinet]

Hmm I see may as well get some 3m xlr cables then.

Do you have audible noise? If so there may be cheaper ways to eliminate it than going for xlr. Start with asking if you need your DAC to be 3m away from your amp.

 

[redjr]

Hi.

Connections via XLR are always better than those via RCA for several reasons -> GOOGLE is your friend (or ask an IA)

You only gain about 3-4dB less noise. When that's over 120dB, you won't hear the difference - even if golden ears. It's especially irrelevant with short runs.

 

[Yviena]

Do you have audible noise? If so there may be cheaper ways to eliminate it than going for xlr. Start with asking if you need your DAC to be 3m away from your amp.

I could probably shorten the run from the dac but that means i have to rearrange more stuff to keep it all nice and tidy in my media rack.

 

[antcollinet]

You only gain about 3-4dB less noise. When that's over 120dB, you won't hear the difference - even if golden ears. It's especially irrelevant with short runs.

Actually no - you are talking about the gain in SNR - simply resulting from the higher voltage.

When there is ground noise/common mode noise present in the system, the balanced signalling ability to reject that will result in 40dB or more reduction, depending on the tolerance of the input and output impedances. This can easily take noise from very audible to inaudible.

 

[antcollinet]

I could probably shorten the run from the dac but that means i have to rearrange more stuff to keep it all nice and tidy in my media rack.

What is driving your DAC? You might be able to break a ground loop (if that is what is causing your noise) by using a toslink connection between source and DAC rather than USB or Coax. If your source is a PC, these are particularly likely to be causing ground noise issues.

It sort of depends where the noise is coming from. Does the noise go away or reduce if you disconnect the source from the DAC input? (Disconnect it completely from the system)

 

[Kpaw]

Can anyone confirm if the USB port is powered ? I’d like to DIY an amp trigger.

 

[Yviena]

What is driving your DAC? You might be able to break a ground loop (if that is what is causing your noise) by using a toslink connection between source and DAC rather than USB or Coax. If your source is a PC, these are particularly likely to be causing ground noise issues.

It sort of depends where the noise is coming from. Does the noise go away or reduce if you disconnect the source from the DAC input? (Disconnect it completely from the system)

It goes away when disconnected from my pc, but I did get a long enough xlr cable and it did the job.

 

[tccalvin]

Can anyone confirm if the USB port is powered ? I’d like to DIY an amp trigger.

Likely not.