SMSL D200 DAC Review

[zdykstra]

There's some good info in this thread. In practice, for modern digital volume control implementations there's no reason to worry about it in the majority of cases.

I would think about it this way: What is the main advantage of analog volume control? Everything in the signal is scaled down, including the DAC's self-noise. That noise sets a hard limit on your SNR and the usable dynamic range, although in most cases the recording will limit it further. So if you dont hear any noise right now using the SU-10 while no music is playing (and the DAC didn't mute automatically), your practical SNR cannot improve by further reducing that already inaudible self-noise. If that is the case, your SNR is limited by either your hearing, the noise floor of your room or the recording.

In contrast, if you can hear noise (usually tweeter hiss) right now, there may be something to gain. But you would need to check whether the culprit is the DAC's or the amp's self-noise. If it's the latter, changing the DAC volume control scheme would also provide no benefit.

As you would expect, there's no audible hiss when there's no music playing. My Dynaudio Contour 60's aren't egregiously sensitive and the DAC and amp are about as perfect as modern tech can get.

Based on what I've read, assuming my input to the DAC is a 16 bit PCM stream that's actually using all 16 bits of range, I could get to -37db of attenuation before the SU-10's noise floor starts being _technically_ part of the output signal going to the amplifier. My wife listens to this system in the morning at about -40 to -60, and I get into the mid -30's in the evening. I guess my comfort hinges on the general dynamic range of most of my CD rips; if they're around 13 bits of dynamic range, I'm not really losing any precision at my listening levels. If they're higher, then it's my understanding that I would technically give up some dynamic range at lower listening levels with digital volume control that operates as a 32bit float internally. Is that correct?

 

[Panelhead]

One step down when using the D200 is volume resolution. Most digital attenuation done inside a dac has 0.5dB steps. The digitally controlled analog volume in the D200 has 2dB steps.
I used to think that is fine. But after using the digital volume controls the 2dB is too coarse.

 

[RandomEar]

My wife listens to this system in the morning at about -40 to -60, and I get into the mid -30's in the evening. I guess my comfort hinges on the general dynamic range of most of my CD rips; if they're around 13 bits of dynamic range, I'm not really losing any precision at my listening levels. If they're higher, then it's my understanding that I would technically give up some dynamic range at lower listening levels with digital volume control that operates as a 32bit float internally. Is that correct?

I think there still would be no audible loss at the lower levels if the self-noise is inauble. Because you need to take into account the actually "usable" part of those 13 bits in your listening situation. That depends on the SPL you end up with in your room. Assuming 0 dB_SPL is your lower hearing limit and the room noise floor isn't relevant (which it is in practice and will probably "gift" you another 10 or 15 dB inauble range for noise), 13 bits gives you about 80 dB of dynamic range. If the peak SPL with your setup ends up below 80 dB at whatever DAC attenuation level, you should definitely be good because the audible DR in your room in that situation would be lower than what your system is delivering.

If I'm not mistaken, attenuatting any further reduces the playback system DR to the same degree as the available DR in the room. That's because the room noise floor is (assumed) constant and the same is true for the DAC noise floor when using digital attenuation. So attenuating more should always be safe as it does not increase the noise level. At the same time, listening at much higher levels could actually exceed one of the DR limits - but most likely the one of the recording. It depends on what is reached sooner: The dynamic range of the recording or that level of available DR at the required attenuation. I fear it is not trivial to answer this universally

In essence, there are multiple factors limiting the required SNR of your system to still sound transparent in a given situation and you would need to determine which is the actually limiting one if you're seeking a universal answer to "is this attenuation method good enough". I still don't think this is worth worrying about too much, though. Unless you notice any immediate problem with audible noise, you should be good.

 

[Roland68]

One step down when using the D200 is volume resolution. Most digital attenuation done inside a dac has 0.5dB steps. The digitally controlled analog volume in the D200 has 2dB steps.
I used to think that is fine. But after using the digital volume controls the 2dB is too coarse.

Is that just a theoretical opinion, or do you actually own a D200 and find that point bothersome when using it?

 

[zdykstra]

I think there still would be no audible loss at the lower levels if the self-noise is inauble. Because you need to take into account the actually "usable" part of those 13 bits in your listening situation. That depends on the SPL you end up with in your room. Assuming 0 dB_SPL is your lower hearing limit and the room noise floor isn't relevant (which it is in practice and will probably "gift" you another 10 or 15 dB inauble range for noise), 13 bits gives you about 80 dB of dynamic range. If the peak SPL with your setup ends up below 80 dB at whatever DAC attenuation level, you should definitely be good because the audible DR in your room in that situation would be lower than what your system is delivering.

If I'm not mistaken, attenuatting any further reduces the playback system DR to the same degree as the available DR in the room. That's because the room noise floor is (assumed) constant and the same is true for the DAC noise floor when using digital attenuation. So attenuating more should always be safe as it does not increase the noise level. At the same time, listening at much higher levels could actually exceed one of the DR limits - but most likely the one of the recording. It depends on what is reached sooner: The dynamic range of the recording or that level of available DR at the required attenuation. I fear it is not trivial to answer this universally

In essence, there are multiple factors limiting the required SNR of your system to still sound transparent in a given situation and you would need to determine which is the actually limiting one if you're seeking a universal answer to "is this attenuation method good enough". I still don't think this is worth worrying about too much, though. Unless you notice any immediate problem with audible noise, you should be good.

Thanks for that detailed reply, I hadn't considered a few of the points you raised. I did just make a wild ass guess on my files having 13 bits of resolution, I've never really run any analysis to track that album by album. I'm sure I'm also conflating some terms here, so take whatever I say with a big grain of salt.

Last night, I switched the gain on the Buckeye amplifier to low. The net effect is that I have to apply less digital attenuation in the DAC - by about 5db or so. There's still more than enough headroom to get louder than I can tolerate in that space, so that seems like the best compromise. And as you noted, I can't actually tell any real difference so I don't think this is worth pursuing that actively.

I have debated picking up another D900, the relay-based volume control is quite nice and the remote is leagues and leagues ahead of the SMSL remote. Maybe if I come across another killer deal...

 

[Panelhead]

Is that just a theoretical opinion, or do you actually own a D200 and find that point bothersome when using it?

No, never seen a D200. But over the last few years have used many preamps with analog volume controls. The better offered 1dB increments. But most have larger steps. Built series and shunt attenuators with 31 steps. Seemed fine 30 years ago.
And for me it is not just 2dB steps. Sometimes I dial in -14.5dB attenuation. Bump up volume to -13.5dB. A digitally controlled 31 step analog attenuator seems incomplete.
There are digitally controlled analog attenuators that offer 0.5dB steps. The MAS6116 offers 128 steps @0.5dB. Even the ancient PGA2311 offers 0.5dB steps. The distortion of the old PGA2311 should be inaudible.
The NJU72315 has a slightly lower SNR compared to a MAS6116. The MAS6116 has four times as many steps.
I really wanted to get a D200. See if there is magic in the ROHM chip.
Two things that stopped me, the lack of digital attenuation in the dac chip, and no trigger.
The two dacs I now use have ES9039 and AK4499 chips. The volume controllers are above transparent. Mainly use the dac with AK4499. It has trigger.

 

[zdykstra]

No, never seen a D200. But over the last few years have used many preamps with analog volume controls. The better offered 1dB increments. But most have larger steps. Built series and shunt attenuators with 31 steps. Seemed fine 30 years ago.
And for me it is not just 2dB steps. Sometimes I dial in -14.5dB attenuation. Bump up volume to -13.5dB. A digitally controlled 31 step analog attenuator seems incomplete.
There are digitally controlled analog attenuators that offer 0.5dB steps. The MAS6116 offers 128 steps @0.5dB. Even the ancient PGA2311 offers 0.5dB steps. The distortion of the old PGA2311 should be inaudible.
The NJU72315 has a slightly lower SNR compared to a MAS6116. The MAS6116 has four times as many steps.
I really wanted to get a D200. See if there is magic in the ROHM chip.
Two things that stopped me, the lack of digital attenuation in the dac chip, and no trigger.
The two dacs I now use have ES9039 and AK4499 chips. The volume controllers are above transparent. Mainly use the dac with AK4499. It has trigger.

https://www.smsl-audio.com/portal/product/detail/id/921.html There are multiple pictures on that page showing -50.5 db . Unless they decided to shift all volume positions by half a dB, the D200 offers half dB volume control.

 

[Roland68]

No, never seen a D200. But over the last few years have used many preamps with analog volume controls. The better offered 1dB increments. But most have larger steps. Built series and shunt attenuators with 31 steps. Seemed fine 30 years ago.
And for me it is not just 2dB steps. Sometimes I dial in -14.5dB attenuation. Bump up volume to -13.5dB. A digitally controlled 31 step analog attenuator seems incomplete.
There are digitally controlled analog attenuators that offer 0.5dB steps. The MAS6116 offers 128 steps @0.5dB. Even the ancient PGA2311 offers 0.5dB steps. The distortion of the old PGA2311 should be inaudible.
The NJU72315 has a slightly lower SNR compared to a MAS6116. The MAS6116 has four times as many steps.
I really wanted to get a D200. See if there is magic in the ROHM chip.
Two things that stopped me, the lack of digital attenuation in the dac chip, and no trigger.
The two dacs I now use have ES9039 and AK4499 chips. The volume controllers are above transparent. Mainly use the dac with AK4499. It has trigger.

That’s what I thought.
The D200 adjusts in 0.5 dB increments and has 120 steps.
SMSL apparently uses a hybrid design here: it always handles the 0.5, 1.0, and 1.5 dB steps digitally, then—at every 2.0 dB step—resets the digital value to 0 dB and shifts the NJU72315 by a 2 dB step. Then the cycle starts all over again.
Actually, it’s quite clever.

 

[Panelhead]

That is clever. It is not a digitally controlled analog attenuator. It is a digitally controlled digital/analog attenuator.
The NJU72315 is definitely offers 31 steps in 2dB increments. Once you have used a attenuator that has 0.5dB steps it would be hard to go back.

 

[zdykstra]

That’s what I thought.
The D200 adjusts in 0.5 dB increments and has 120 steps.
SMSL apparently uses a hybrid design here: it always handles the 0.5, 1.0, and 1.5 dB steps digitally, then—at every 2.0 dB step—resets the digital value to 0 dB and shifts the NJU72315 by a 2 dB step. Then the cycle starts all over again.
Actually, it’s quite clever.

It'd be great to see that technique adopted in more DACs that are intended to directly drive an amplifier. Short of an actual dedicated preamplifier circuit, that's about as good as you can hope for.

 

[Roland68]

That is clever. It is not a digitally controlled analog attenuator. It is a digitally controlled digital/analog attenuator.
The NJU72315 is definitely offers 31 steps in 2dB increments. Once you have used a attenuator that has 0.5dB steps it would be hard to go back.

It is a purely analog resistor network controlled digitally via an I2C bus—essentially, purely analog attenuation.

As mentioned, in the D200, the NJU72315 is configured for 120 steps, resulting in 0.5 dB increments. Theoretically, this would also be possible with any other DAC chip featuring digital volume control.

Alternatively, there is the slightly more expensive Muses series: the NJU72320/NJU72322 with 0.5 dB steps, and the NJU72323 with 0.25 dB steps.

 

[Panelhead]

It is a two bit digital attenuator combined with the NJU72315.

 

[piaseczek]

It is a two bit digital attenuator combined with the NJU72315.

Exactly, vmv d3r uses the same solution.

 

[Roland68]

It is a two bit digital attenuator combined with the NJU72315.

Yes, but I already described the principle behind it in post #248.

 

[Bubblefish]

Hello everyone,

I started using a current version of the SMSL D200 DAC yesterday and joined this forum to connect with fellow enthusiasts. I would also like to share my experience and hope to get some feedback/suggestions.

I purchased the device because of its "digital-analog" volume control, its "musical/dynamic" sound profile, and, of course, its currently low price.
The rest of my chain consists of a WiiM Pro Plus connected to the D200 via optical SPDIF, a TV connected to the WiiM via optical SPDIF, and DIY speakers (featuring a coaxial midrange/tweeter setup) powered by a TPA3255 amplifier with analog input control (no remote).

Volume Control:
Perhaps I missed the relevant explanations in this forum, but according to the display, the volume control starts at -0 dB and "ends" at -59.5 dB with MUTE. The volume gets muted completely, and if the amplifier's input is set to full output, there is a more or less distinct jump in volume. If I want to finely adjust the volume all the way down to a whisper, I either have to use the amplifier's analog control or resort to digital manipulation in the WiiM Pro Plus. Therefore, this hybrid solution is only clever down to -59.5 dB.

Dynamics:
Either I am no longer used to this kind of dynamic range in music, or I am confusing dynamics with a malfunction of the DAC. Across all genres I've listened to in the first couple of hours, the music seems to rhythmically swell and fade. It feels and sounds like a "pumping" effect. I cannot say for sure, when heard, whether it is rhythmically driven by the music itself or independent of it, like a cyclical pumping.

Interestingly, I had to deactivate the PEQ in the WiiM Pro Plus — which previously acted as my DAC — which I had been using to compensate for room acoustics. Doing so made the new soundstage significantly more open, wider, deeper, more layered, and more precise. The spontaneous comparison that comes to mind is like driving a car and finally releasing the handbrake.

Thanks
Bubblefish

 

[tccalvin]

Either I am no longer used to this kind of dynamic range in music, or I am confusing dynamics with a malfunction of the DAC. Across all genres I've listened to in the first couple of hours, the music seems to rhythmically swell and fade. It feels and sounds like a "pumping" effect. I cannot say for sure, when heard, whether it is rhythmically driven by the music itself or independent of it, like a cyclical pumping.

One thing is for sure, switching to a new DAC will not expand the dynamic range of your music, even if said DAC has more dynamic range than your previous one. What you might notice is a slightly quieter noise floor, but that's it (unless you're coming from a 16-bit DAC...).

Interestingly, I had to deactivate the PEQ in the WiiM Pro Plus — which previously acted as my DAC — which I had been using to compensate for room acoustics. Doing so made the new soundstage significantly more open, wider, deeper, more layered, and more precise. The spontaneous comparison that comes to mind is like driving a car and finally releasing the handbrake.

If you enjoy your setup more without EQ that's all well and good, but the D200 probably made no difference in the frequency response of your setup. The WiiM's frequency response is flat as a board:

 

[Bubblefish]

One thing is for sure, switching to a new DAC will not expand the dynamic range of your music, even if said DAC has more dynamic range than your previous one. What you might notice is a slightly quieter noise floor, but that's it (unless you're coming from a 16-bit DAC...).

I’m still not sure whether the relatively small difference in the noise floor between the two DACs (AKM vs. Rohm) is sufficient to produce this effect.

Perhaps some kind of dynamic expansion is also playing a role here, creating the impression that more is happening in the music than is actually the case.
It doesn’t have to be much and could look like the following PEQ filter:
PK: 180 Hz, -1,5 dB, Q 0,9
PK: 350 Hz, -1,0 dB, Q 1,0
PK: 2,5 kHz, +0,8 dB, Q 0,8
HS: 10 kHz, +0,8 dB
Preamp: -2,5 dB

However, I can’t identify this effect in the measurements. Still, if that is indeed what’s happening, it would be great to have the option to disable this expansion.

If you enjoy your setup more without EQ that's all well and good, but the D200 probably made no difference in the frequency response of your setup. The WiiM's frequency response is flat as a board:

View attachment 557945

The signal is corrected in the WiiM and reaches the D200 with the room correction already applied. I’m trying to imagine that the dynamics might be related to this: if there are higher levels at the lower and upper ends of the frequency spectrum, and these roughly correspond to the PEQ correction, then they would now be present twice. In that case, one of them would have to be removed.

 

[Roland68]

I’m still not sure whether the relatively small difference in the noise floor between the two DACs (AKM vs. Rohm) is sufficient to produce this effect.

Perhaps some kind of dynamic expansion is also playing a role here, creating the impression that more is happening in the music than is actually the case.
It doesn’t have to be much and could look like the following PEQ filter:
PK: 180 Hz, -1,5 dB, Q 0,9
PK: 350 Hz, -1,0 dB, Q 1,0
PK: 2,5 kHz, +0,8 dB, Q 0,8
HS: 10 kHz, +0,8 dB
Preamp: -2,5 dB

However, I can’t identify this effect in the measurements. Still, if that is indeed what’s happening, it would be great to have the option to disable this expansion.

The signal is corrected in the WiiM and reaches the D200 with the room correction already applied. I’m trying to imagine that the dynamics might be related to this: if there are higher levels at the lower and upper ends of the frequency spectrum, and these roughly correspond to the PEQ correction, then they would now be present twice. In that case, one of them would have to be removed.

Neither the D200 nor the installed ROHM BD34352EKV chip possesses any functionality that would allow or cause such a thing.

As a first step, you could try performing a factory reset on the D200. Be careful with the volume level afterwards.

To rule out external factors, please connect a different source—for example, a PC/Mac via USB or a CD player via SPDIF.

Have you connected both speakers with the correct polarity (in phase)?
Are there any issues with your audio cables connecting to the amplifier? Check for phase errors, cold solder joints, or significant differences in resistance.

 

[Panelhead]

Also visit the gain staging of your system. There is too much gain available if max attenuation is still higher than a desired volume.
The issue may be in your amplifier. It is unusual that a pair of speakers (90dB efficient) and an amplifier (maybe 23dB voltage gain) is too loud for some listening with -60dB dropped in the dac. It should be a whisper.
I have extremely high efficiency speakers and have never needed to pad down a dac that much. But have used mid 40’s. That was with an amplifier with normal voltage gain.
I now use the dac chips to control volume. And an amplifier with zero voltage gain. This way dac is padded down -10 to -20dB depending on target volume and level of the track.
Using the D200 at even -40dB is hurting the overall sonics. The signal is being boosted to a high level somewhere and then reduced a 100 times to get a desired output level.

 

[Bubblefish]

Neither the D200 nor the installed ROHM BD34352EKV chip possesses any functionality that would allow or cause such a thing.

As a first step, you could try performing a factory reset on the D200. Be careful with the volume level afterwards.

To rule out external factors, please connect a different source—for example, a PC/Mac via USB or a CD player via SPDIF.

Have you connected both speakers with the correct polarity (in phase)?
Are there any issues with your audio cables connecting to the amplifier? Check for phase errors, cold solder joints, or significant differences in resistance.

It took a while, but I was finally able to follow the recommendation after purchasing a new coaxial cable.

Thank you for the tip: I can hardly believe that resetting a brand-new D200 to factory settings and using a coaxial cable solved the issue, but I followed the suggested steps. After the reset and a flawless USB test with my FiiO M21, I connected the WiiM via both coaxial and optical cables and was able to identify the faulty optical cable in a direct comparison.

The speakers were already correctly wired to the amplifier.

Thanks, Bubblefish