Review and Measurements of Topping D50 DAC

[typericey]

Thanks for the filter responses Amir. Can we conclude that the filters have no audible difference because all the differences occur beyond the range of hearing?

Btw, happy to report that I was able to install the latest firmware without a hitch.

 

[yue]

Thanks for the filter responses Amir. Can we conclude that the filters have no audible difference because all the differences occur beyond the range of hearing?

Btw, happy to report that I was able to install the latest firmware without a hitch.

If you're older than 20 then definately. Otherwise a few filters roll off too quickly before 20khz, and it's possible young people may hear a difference.

 

[yue]

Btw, happy to report that I was able to install the latest firmware without a hitch.

I'm still talking with Topping and guide them solving a few firmware issues, notably
- the firmware does not broadcast 16 bits playback capability on all platforms(on Linux 24 bits playback is also missing)
- There's a channel naming issue that causes Mac OS X and Linux (perhaps Windows?) not able to adjust volume correctly.
Hopefully they can provide an update to the firmware again.

 

[Vosya]

There was a request for the filter responses of Topping D50. Here they are:

The situation is depressing if we compare the results with the measurements of Topping DX7s. Not to mention the declared values of es9038q2m. I am sincerely disappointed with the Topping D50.
P.S. How, in your opinion, can be explained by 40-50 dB the worst value of the stop band in Topping D50? And could you further take measurements for 48 kHz for comparison?

 

[amirm]

The situation is depressing if we compare the results with the measurements of Topping DX7s. Not to mention the declared values of es9038q2m. I am sincerely disappointed with the Topping D50.
P.S. How, in your opinion, can be explained by 40-50 dB the worst value of the stop band in Topping D50? And could you further take measurements for 48 kHz for comparison?

I don't know. I should double-check the results.

 

[miero]

These are caused by a high volume of "white noise" signal that overloads the FIR of the DAC. The same effect happens with ES9018. I wrote about it today in https://www.audiosciencereview.com/...-soekris-dac1421-multibit-dac.3956/post-94476

Such signal contains intersample peaks (higher than 0dBFS) that will be generated by DAC by its internal (e.g. 8x) oversampling, and the FIR filter that follows after oversampling will not properly limit it.

This can be simulated also in a digital domain using a SoX tool, for example on a randomly generated white noise of 60 seconds duration:
$ sox -b 32 -r 44100 -n -n synth 60 white rate $((8*44100)) sox WARN rate: rate clipped 33210 samples; decrease volume?

It helps to decrease a volume of such signal, for example by 6dB so it will not clip after internal oversample in the DAC:
$ sox -b 32 -r 44100 -n -n synth 60 white vol -6dB rate $((8*44100)) stats ... Pk lev dB -0.06 RMS lev dB -10.98 ...

 

[Vosya]

These are caused by a high volume of "white noise" signal that overloads the FIR of the DAC.

Shown in the discussed graph -30 dB is a "high" level? Аnd why under the same conditions DX7s shows an almost ideal graph?

 

[miero]

The volume level -30dB shown in the graph is just an average level in a single bin of FFT. You need to sum volume in all bins to get the final average volume level.

A volume of a white noise bellow 24kHz in the following graph is approx. -36dB + 39dB (= 10*log(23000/96000*32768)/log(10)) = +3dBFS (notice, there is a red value stating RMS +3dB).

FIY:
2 equal loud sound sources -> +3dB
4 equal loud sound sources -> +6dB
8 equal loud sound sources -> +9dB
16 equal loud sound sources -> +12dB
32 equal loud sound sources -> +15dB
64 equal loud sound sources -> +18dB
128 equal loud sound sources -> +21dB
256 equal loud sound sources -> +24dB
512 equal loud sound sources -> +27dB
1024 equal loud sound sources -> +30dB
2048 equal loud sound sources -> +33dB
...
More info: http://www.sengpielaudio.com/calculator-leveladding.htm

 

[Vosya]

The next question about DX7s was not noticed? Or is there no answer to this simple question?

 

[miero]

It would help a lot if you provide a link to image you mentions...

 

[Vosya]

I thought that there was a not abstract discussion...
https://audiosciencereview.com/foru...pping-dx7s-dac-and-headphone-amp.2378/page-17

 

[n2it]

@miero this post https://www.audiosciencereview.com/...ng-dx7s-dac-and-headphone-amp.2378/post-92656

 

[miero]

OK, I see.

I guess that the firmware of DX7s sets value of "master trim level" on ES9038Q2M to smaller value than the firmware of D50. And that the response D50 filters can be "fixed" by attenuating its volume by 3-6dB.

 

[Vosya]

I guess that the firmware of DX7s sets value of "master trim level" on ES9038Q2M to smaller value

According to the es9038q2m structural scheme, volume control is located before the FIR filter?

 

[LittleNemo]

I'm still talking with Topping and guide them solving a few firmware issues, notably
- the firmware does not broadcast 16 bits playback capability on all platforms(on Linux 24 bits playback is also missing)
- There's a channel naming issue that causes Mac OS X and Linux (perhaps Windows?) not able to adjust volume correctly.
Hopefully they can provide an update to the firmware again.

So this is an issue? I thought that it just upsampled 16bit to 24bit.

 

[yue]

So this is an issue? I thought that it just upsampled 16bit to 24bit.

missing 24 bit means no dop on Linux.
extending 16 to 24 or 32 bit means higher CPU/battery usage while getting no benefit at all.
it also crashes certain apps (such as BitPerfect on Mac) as reported in this forum.

 

[LittleNemo]

Yea I know. What I'm saying is that I'm playing 16/44.1 FLAC via usb over Rpi3b and and output on the Topping is showing 32bit

 

[Toku]

Is the Bit number displayed on the D50 display a problem?

The bit display on the display will be 32 bits, the maximum bit rate of the USB connection.
I understand like that. Even DX 7s is displayed as 32 Bit.

 

[yue]

Yea I know. What I'm saying is that I'm playing 16/44.1 FLAC via usb over Rpi3b and and output on the Topping is showing 32bit

right it won't hurt you if you stick with PCM, apart from some performance cost and battery drain (esp. on raspberry pi), as there're quite a lot of bit manipulation going on in your raspberry pi. But if you do DSD it's another story ---

Is the Bit number displayed on the D50 display a problem?

The bit display on the display will be 32 bits, the maximum bit rate of the USB connection.
I understand like that. Even DX 7s is displayed as 32 Bit.

If you use Linux the only supported bit depth option is 32. I suggest them to fix it because it sounds like a bug to me rather than a feature.
It also crashes certain players with up-depths feature, or make them no longer working.

 

[ramzi]

Does anyone have a way of confirming the D50 is running the new firmware? From what I can see, after flashing the new firmware, the version is still displayed as v6.F2 which was the initial version if I'm not mistaken...