SMSL D400 Pro Balanced DAC Review

[ClaudeJ1]

"Modern Audio is a Quasi-Religious experience. Of course the more expensive one will always sound better that the cheap one. Despite high resolution measurements to the contrary. My friend of 52 years will never be convinced otherwise, despite the FACTS."

Very true. Sadly so...

Yes. What I admire about this place is the absence of that Quasi Religious "emotional testing" so prevalent elsewhere on the Web.

 

[Roland68]

OK, do your really need the update when the other 99% of the functions are so good? The ignoring the 99% good performance of this device and since it cannot please your specific application, you should do it. This is like the people who complained about the issues of the Topping PA-5 (I never had an overheat problem powering HORNS), fixed in version II and they have never been turned off for over a year. Likely 'cause everyone uses too many watts instead of getting more efficient speakers to play louder. Fix the root cause, not bitch about great products that cost almost nothing compared to stuff even 20 years ago. If it can't be fixed to your specific liking, then, get something else.

But people who spread fairy tales about devices are also a problem.
The PA5 never had an overheating problem, it wasn't the power supply or inefficient speakers or any other nonsense.
The source of the problem was proven to be the potting compound that Topping used improperly and incorrectly to protect their "secrets". This potting compound led to the failures and nothing else.
I have now repaired over 10 PA5s without having to replace a single OPAmp or other components, so much for the overheating problem.
The problem was improper use of the potting compound, nothing more and nothing less.

If you buy a car and 99 things work great but the reverse gear doesn't, then you don't complain?
Or if 99 things work great with your AVR but the left front speaker doesn't, then that's OK for you too?
Every buyer has the right to decide for themselves whether a function or guaranteed feature of a device is important to them or not.

And SMSL has really bad support when it comes to problems/errors caused by firmware. And why are such simple errors not discovered beforehand in internal tests.
How many devices currently have firmware problems, 5, 10 or 20? And not just for days or weeks, but some for months or years. Problems with filters, channel assignments, swapped channels for some functions, various problems with I2S, etc. Devices with faulty firmware have also been resold, even though it was clear that customers had to send these devices in for firmware updates. When I asked how I could get a device with the current firmware, I was told that I should either wait until the devices with the old firmware were sold out or order from a specific dealer in China. Very professional.

 

[Kiko1974]

But people who spread fairy tales about devices are also a problem.
The PA5 never had an overheating problem, it wasn't the power supply or inefficient speakers or any other nonsense.
The source of the problem was proven to be the potting compound that Topping used improperly and incorrectly to protect their "secrets". This potting compound led to the failures and nothing else.
I have now repaired over 10 PA5s without having to replace a single OPAmp or other components, so much for the overheating problem.
The problem was improper use of the potting compound, nothing more and nothing less.

If you buy a car and 99 things work great but the reverse gear doesn't, then you don't complain?
Or if 99 things work great with your AVR but the left front speaker doesn't, then that's OK for you too?
Every buyer has the right to decide for themselves whether a function or guaranteed feature of a device is important to them or not.

And SMSL has really bad support when it comes to problems/errors caused by firmware. And why are such simple errors not discovered beforehand in internal tests.
How many devices currently have firmware problems, 5, 10 or 20? And not just for days or weeks, but some for months or years. Problems with filters, channel assignments, swapped channels for some functions, various problems with I2S, etc. Devices with faulty firmware have also been resold, even though it was clear that customers had to send these devices in for firmware updates. When I asked how I could get a device with the current firmware, I was told that I should either wait until the devices with the old firmware were sold out or order from a specific dealer in China. Very professional.

I couldn't have explained It better.
I don't know why brands like Topping or SMSL, that in general make great sounding gear, don't properly check their devices, firmware included, before being sold.

 

[Roland68]

I couldn't have explained It better.
I don't know why brands like Topping or SMSL, that in general make great sounding gear, don't properly check their devices, firmware included, before being sold.

It's quite simple, it's a costly and time-consuming process between pure hardware development and the start of production/sales.
This requires skilled people, equipment and time, all things that increase costs and delay market launch.

It's much cheaper to use end customers as beta testers, which wouldn't be so bad for pure software problems if solutions were available in the form of firmware updates in a timely manner. But that's exactly what doesn't work in most cases, especially with SMSL, Loxjie and Sabaj.
How many months has the filter problem been around with ES9039 DACs? And it's just about something as simple as incorrect assignment of the order in the menu.
The DO400 has been on the market for 2 years with this error (and others).
How much more ridiculous can it get?

Of course, no manufacturer wants to invest money in delivered devices, but that can't be the end customer's problem.
For the end customer, non-functioning software has the same status as a hardware defect in a device until it is fixed. That's why it's an absolutely stupid idea for a customer to settle for a defect, even if the rest works. Selling a used device with a defect will always mean a financial loss for the original buyer.
As an end customer, you only have the option of putting the dealer and manufacturer in default and returning the device within the statutory warranty period.
If enough end users do that, something could change.

 

[Kiko1974]

It's quite simple, it's a costly and time-consuming process between pure hardware development and the start of production/sales.
This requires skilled people, equipment and time, all things that increase costs and delay market launch.

It's much cheaper to use end customers as beta testers, which wouldn't be so bad for pure software problems if solutions were available in the form of firmware updates in a timely manner. But that's exactly what doesn't work in most cases, especially with SMSL, Loxjie and Sabaj.
How many months has the filter problem been around with ES9039 DACs? And it's just about something as simple as incorrect assignment of the order in the menu.
The DO400 has been on the market for 2 years with this error (and others).
How much more ridiculous can it get?

Of course, no manufacturer wants to invest money in delivered devices, but that can't be the end customer's problem.
For the end customer, non-functioning software has the same status as a hardware defect in a device until it is fixed. That's why it's an absolutely stupid idea for a customer to settle for a defect, even if the rest works. Selling a used device with a defect will always mean a financial loss for the original buyer.
As an end customer, you only have the option of putting the dealer and manufacturer in default and returning the device within the statutory warranty period.
If enough end users do that, something could change.

Maybe brands like SMSL and others shouldn't release so many different products on the same category, DACs being one example, and make sure that the products work properly, and don't feature configuration options that don't actually work, in the case of the D400 PRO, the I2S channel invertion when playing DSD that doesn't work, despite the so called firmware update that they sent me via e-mail around 3 months ago, that by the way IS the same they have available on their website for a few weeks, and doesn't work. There's no way to do the firmware update despite following the provided instructions verbatim.
As talking end users as beta testers for a new product, It sounds more like we are trested more like Guinea pigs.

 

[Kiko1974]

One thing that is not clear to me is on the Pre Mode and how DSD is trested/processed.
There are two options, one is "Fixed" and a second one that is "Fixed DSD Bypass".
Being this DAC a two IC's solution, AK4191 doing oversampling and Delta Sigma modulation, and the AK4499EX the actual D/A IC, what kind of processing is being done to DSD when the "Fixed" Pre Mode IS selected? Is It converted to 7 bit Delta Sigma modulation? And when the "Fixed DSD Bypass"?

 

[Scytales]

Just read the SMSL D400 user manual and compare the claims about those features to the AK4191 datasheet.

I think the "Fixed" pre-mode option just means that the on-chip digital volume control in the AK4191 DSP is disable, whereas the "Fixed DSD Bypass" is the implementation of the DSD bypass mode in the AK4191, ie DSD signal does not pass through the DSD digital volume control processor, neither the following multilevel sigma-delta modulator, only through a digital low-pass filter to tame the DSD HF quantization noise. The resulting multibit data produced by this filter is then directly sent to the following D/A converter chip.

When the "Fixed" pre-mode is selected, it must mean that DSD signals pass through a digital low-pass filter, then the DSD digital attenuator (although at a fixed setting because volume control is deactivated), then the sigma-delta modulator.

The publicly available block diagram of the AK4191 is straightforward about that matter:

 

[Kiko1974]

Just read the SMSL D400 user manual and compare the claims about those features to the AK4191 datasheet.

I think the "Fixed" pre-mode option just means that the on-chip digital volume control in the AK4191 DSP is disable, whereas the "Fixed DSD Bypass" is the implementation of the DSD bypass mode in the AK4191, ie DSD signal does not pass through the DSD digital volume control processor, neither the following multilevel sigma-delta modulator, only through a digital low-pass filter to tame the DSD HF quantization noise. The resulting multibit data produced by this filter is then directly sent to the following D/A converter chip.

When the "Fixed" pre-mode is selected, it must mean that DSD signals pass through a digital low-pass filter, then the DSD digital attenuator (although at a fixed setting because volume control is deactivated), then the sigma-delta modulator.

The publicly available block diagram of the AK4191 is straightforward about that matter:

 

[Kiko1974]

I don't quite understand what's going on when the Fixed DSD Bypass mode is on.
I get that the DSD Signal IS not going through any kind of volume control, but I throught that when this Mode is enabled, It completely bypasses the AK4191 oversampling/Delta Sigma modulator, yet you say that when Fixed DSD Bypass is on the incoming 1 bit DSD is converted to 7 bit Delta Sigma modulation. To do that it's MUST PASS through the AK4191 IC.
So my understanding IS that when Fixed DSD Bypass is on, DSD isn't actually converted as 1 bit at 2.8 Mhz, its native resolution.
Also, there's the fact that the so called Sound Colors implementes on this two ICs solution, are non detestable. Sound Color 1& 2 makes the AK4499EX work at 5.6 Mhz, that's double the rate of SACD DSD, and when Sound Color 3 & 4 are on, the make the AK4499EX work at 11.2 Mhz, that's DSD',s sampling rate x2 and x4.

 

[Scytales]

I don't quite understand what's going on when the Fixed DSD Bypass mode is on.
I get that the DSD Signal IS not going through any kind of volume control, but I throught that when this Mode is enabled, It completely bypasses the AK4191 oversampling/Delta Sigma modulator, yet you say that when Fixed DSD Bypass is on the incoming 1 bit DSD is converted to 7 bit Delta Sigma modulation. To do that it's MUST PASS through the AK4191 IC.
So my understanding IS that when Fixed DSD Bypass is on, DSD isn't actually converted as 1 bit at 2.8 Mhz, its native resolution.
Also, there's the fact that the so called Sound Colors implementes on this two ICs solution, are non detestable. Sound Color 1& 2 makes the AK4499EX work at 5.6 Mhz, that's double the rate of SACD DSD, and when Sound Color 3 & 4 are on, the make the AK4499EX work at 11.2 Mhz, that's DSD',s sampling rate x2 and x4.

I didn't wrote that in the "Fixed DSD Bypass" mode, the low-passed filter DSD data stream is remodulated by the sigma-delta modulator. This modulator intervenes in what SMSL names the normal "Fixed" mode.

I have outlined the DSD bypass mode of the AK4191 in red on the block diagram :

To me, it straightforwardly means that in this mod, DSD signals are just passed through a digital low pass filter before reaching the following DAC chip. The digital filter is most probably a FIR type. To my understanding, it is the same data-path the designers of the Burr Brown PCM1738 DAC have described in the document I have cited here. Their drawing makes the principle of operation very clear:

The digital filter produces a multi-bit code at the same sampling frequency (FS) of the input DSD signal (k=h) or higher (k<h) than can be sent directly to the actual D/A converter stage (in the case of the AK4191 DSP, the following AK4499 DAC chip). The corner frequency and shape of attenuation of the low-pass filter can be tweaked by changing the coefficient (a0, a1, a2, etc...) and/or the number of delay lines (D) in actual use, hence the different types of filter available. Yes, the following conversion in not of the one bit type, but a multi-bit conversion working at very high speed (several MHz), same as the digital filter output sample rate.

 

[Kiko1974]

I didn't wrote that in the "Fixed DSD Bypass" mode, the low-passed filter DSD data stream is remodulated by the sigma-delta modulator. This modulator intervenes in what SMSL names the normal "Fixed" mode.

I have outlined the DSD bypass mode of the AK4191 in red on the block diagram :

View attachment 420965

To me, it straightforwardly means that in this mod, DSD signals are just passed through a digital low pass filter before reaching the following DAC chip. The digital filter is most probably a FIR type. To my understanding, it is the same data-path the designers of the Burr Brown PCM1738 DAC have described in the document I have cited here. Their drawing makes the principle of operation very clear:

The digital filter produces a multi-bit code at the same sampling frequency (FS) of the input DSD signal (k=h) or higher (k<h) than can be sent directly to the actual D/A converter stage (in the case of the AK4191 DSP, the following AK4499 DAC chip). The corner frequency and shape of attenuation of the low-pass filter can be tweaked by changing the coefficient (a0, a1, a2, etc...) and/or the number of delay lines (D) in actual use, hence the different types of filter available. Yes, the following conversion in not of the one bit type, but a multi-bit conversion working at very high speed (several MHz), same as the digital filter output sample rate.

Now I get It, thanks for your explanation, very clear and easy to understand.
So what the Fixed DSD Bypass mode does is, well, bypass the volumen control and remodulate/convert the Delta Sigma DSD date from 1 to 7 bits keeping its native Sample rate.

 

[Scytales]

I wouldn't speak about a re-modulation (it isn't), neither a conversion: it is plainly a low pass digital filter whose output contains coded samples which describes the shape of the response in the frequency domain.

 

[Scytales]

I just stumbled across measurements realized with an Audio Precision APX555 of a converter which uses AK4191 DSP and AK4499 DAC that are very interesting to illustrate the points made above.

In this first message, you can see "Amirm-like" dashboards and wide-band FFTs of AK4191+AK4499 in "Bypass" mode with (from top to bottom) DSD64, DSD128, DSD512 and DSD1024 input to the AK4191 : https://audiophilestyle.com/forums/...tal-filtering/?do=findComment&comment=1288738

In "Bypass mode", DSD only passes through the digital filter. The effect of the digital filter can be seen on the wide-band FFTs: the computation produces the nulls on the shaped high frequency noise response.

In this second message, you can see the same measurements in "Normal mode" with (from top to bottom) DSD64, DSD128, DSD512 and DSD1024 : https://audiophilestyle.com/forums/...ss-digital-filtering/page/33/#comment-1288739

In "Normal mode", DSD passes through (another) digital filter, possibly the digital volume control (if enable) and the multilevel delta-sigma modulator before being sent to the DAC chip.

 

[ClaudeJ1]

But people who spread fairy tales about devices are also a problem.
The PA5 never had an overheating problem, it wasn't the power supply or inefficient speakers or any other nonsense.
The source of the problem was proven to be the potting compound that Topping used improperly and incorrectly to protect their "secrets". This potting compound led to the failures and nothing else.
I have now repaired over 10 PA5s without having to replace a single OPAmp or other components, so much for the overheating problem.
The problem was improper use of the potting compound, nothing more and nothing less.

If you buy a car and 99 things work great but the reverse gear doesn't, then you don't complain?
Or if 99 things work great with your AVR but the left front speaker doesn't, then that's OK for you too?
Every buyer has the right to decide for themselves whether a function or guaranteed feature of a device is important to them or not.

And SMSL has really bad support when it comes to problems/errors caused by firmware. And why are such simple errors not discovered beforehand in internal tests.
How many devices currently have firmware problems, 5, 10 or 20? And not just for days or weeks, but some for months or years. Problems with filters, channel assignments, swapped channels for some functions, various problems with I2S, etc. Devices with faulty firmware have also been resold, even though it was clear that customers had to send these devices in for firmware updates. When I asked how I could get a device with the current firmware, I was told that I should either wait until the devices with the old firmware were sold out or order from a specific dealer in China. Very professional.

The "but" word simply wipes out my entire point. While superseding it with your own, which implies it's more important than my original.

 

[Kiko1974]

Let's go back to the so called firmware update for the SMSL D 400 PRO available on SMSL's website, that was sent to me via e-mail at least two months before It was postee on their website, and that I and others, find impossible to update.
I e-mailed SMSL again telling them about their impossible to update new firmware, and with well chosen words, they trested me as if I was retarded and never had done a firmware update before, which I have plenty of time to different devices.
Has anybody been able to do the firmware update to the SMSL D 400 PRO?