Review and Measurements of Topping DX3Pro DAC and Headphone Amp

[kukocz]

The noise in Topping Dx3Pro is divided in two types of noise. One is from outside power supply, aka adapter (we talked about it last time)
and the other one is from inside unit power supply. I studied internal noise yesterday and I found at the RCA output a fundamental frequency of 570khz. After that I was looking at the input power supply votage (15Vdc) with the DAC "on" and I found the same fundamental frequency in the voltage ripple. Here, at the input, the voltage ripple has 120mVp-p ! So, the first voltage regulator from inside the unit has a big charging currents at 570khz. Here is the culprit. I looked to see what regulator it is and I found TPS54331. In the datasheet we found that: TPS54331 it is a step-down DC-DC converter with 570khz switching frequency. It has at the output an LC filter. This capacitor must to be large enough (in Dx3Pro is) and very important, to have low ESR at the frequency of interest. Unfortunately Topping put an electrolytic capacitors here. Until 100khz that type of capacitors could be very good but at 570khz and upper they lose their efficacy. Probably it was better few low-ESR tantalum capacitors in parallel or ceramic high value again in parallel mode. The best would be an linear regulator but the space is too small.

Are you planning to replace this capacitor and re-test? For now, it would be nice to see linearity measurements for switching supply vs lab supply.
I'm wondering if, at the end, it will be worth to change SPS to LPS, replace the capacitor and how much improvement it will bring.

 

[777]

I tried to break it out last night, but I did not get it face down. I need a very long inbus key. Another option could be with a ferrite on RCA output, or a filter, ferrite + capacitor(let's say 330pF).

 

[Toku]

I've combed through this thread, and I still don't understand why there are two seemingly current (but differently numbered) versions of the driver available.

• v4.14 is the latest one with a dedicated entry on Topping's support page (dated 2018-11-20)
• v4.47 doesn't have its own entry, but it's available through a link in the entry for the latest firmware update (dated 2018-12-17)

Can anyone shed some light on this? Maybe it's been covered already and I missed it, but I did look.

Topping's USB driver has V4.14.0 V4.43.0 V4.47.0. V4.47.0 is the latest version. Both drivers work with Topping's DAC D10 D50 DX3 Pro DX 7s. V4.43.0 was taught from Topping. You can download from the following.

Topping_UsbAudio_v 4.43.0 - setup.zip
https://www.dropbox.com/s/0gtfidumxa41e2k/Topping_UsbAudio_v4.43.0-setup.zip?dl=0

 

[Vosya]

Topping_UsbAudio_v 4.43

What is the point of this driver version? In addition to the noticeable corrections made in the new version, ToppingUsbAudioDfu is missing for 4.43, which prevents the firmware from being updated.

 

[splkn]

I tried to break it out last night, but I did not get it face down. I need a very long inbus key. Another option could be with a ferrite on RCA output, or a filter, ferrite + capacitor(let's say 330pF).

Tell me please what is the voltage at the output of TPS54331? And what does it nourish?

 

[pejakovic1]

Will Mean Well GST-25E15-P1J do as a power adapter?

 

[Toku]

What is the point of this driver version? In addition to the noticeable corrections made in the new version, ToppingUsbAudioDfu is missing for 4.43, which prevents the firmware from being updated.

Topping 's USB driver was supplied by XMOS, and XMOS decided Ver number.
There is no explanation about each version, so the difference is unknown.
V4.43.0 was introduced when I was in contact with Topping due to D50's DSD playback problem.

 

[Toku]

Will Mean Well GST-25E15-P1J do as a power adapter?

It is a product of sufficient quality.

 

[777]

Tell me please what is the voltage at the output of TPS54331? And what does it nourish?

I could not measure the voltage at TPS54331 output because I could pull the board out just a two centimeters. I could measure the voltage on a OPA1612. Here we have +/- 9.56v. I tried to add decoupling capacitors on the OPA1612 power supply, 220uF/10v for each voltage rail but the whole device enters in standby mode. There is probably a smps regulator with negative feedback that wants to the output a fixed value of the capacity. As long as I can not undo it, I can not measure and do anything about it. Do you think Topping would give us the schematic of the Dx3Pro ?

In the meantime it stay about 10 hours in ON state, with voltage. The difference in sound is obvious. Most likely the electrolytic capacitors have formed, or better formed. In the first hour the sound was weak, dull, as if it had no current. Now sounds almost perfect.

 

[MetalheadRich]

Since a lot of people were talking about power supply, any idea if running it from a linear power supply like a cheap TeraDak would be worth it?

The noise on this is so low that even a perfect power supply would make no audible difference.

 

[777]

You can not see all the noise with a 90khz bandwidth FFT. You saw all the noise on the oscilloscope and has a aprox. 40mVp-p

 

[peder2tm]

You can not see all the noise with a 90khz bandwidth FFT. You saw all the noise on the oscilloscope and has a aprox. 40mVp-p

Can you do the FFT of the recorded waveform? I would like to see where the noise is in frequency. Which sample frequency did your scope use? If it is possible to see it on the scope we can also determine it's frequency content.

 

[777]

I'll try. My osc have 1 Meg pixel FFT. Never use it but I can try.

 

[splkn]

I tried to break it out last night, but I did not get it face down. I need a very long inbus key. Another option could be with a ferrite on RCA output, or a filter, ferrite + capacitor(let's say 330pF).

If these voltage stabilizers feed the analog part then it is possible. And if it is digital, it can also cause errors as a result of distortion. Therefore, there they need to be filtered immediately.
As for the scheme from topping, I don’t know, at least they should give a detailed flowchart.

 

[yue]

Do you think Topping would give us the schematic of the Dx3Pro ?

I think the best way is to talk to them and diagnose the noise issue together. They are pretty responsive and could probably provide some improvement suggestions.

 

[Vosya]

There is no explanation about each version, so the difference is unknown.

Unknown to those who do not know how to use search engine on the Internet:
-----------------------------------------------------
V4.47.0 (July 31, 2018)
-----------------------------------------------------
* Chg: QPC runtime measurement added again
* Fix: Spy tool doesn't work correctly with virtual channels
* Chg: improved Spy tool statistics
* Chg: API version 5.2
* New: API function TUSBAUDIO_QueryDeviceStatistics added
-----------------------------------------------------
V4.46.0 (July 24, 2018)
-----------------------------------------------------
* New: Workaround for driver unload problems (Code 38 in device manager)
caused by KS layer leaking references on our child PDO.
Registry parameter EnableLeakedPdoRefsWorkaround (true by default)
* New: Added optional PreferDeviceInfo to the KS .inf to use string descriptor of
the audio control interface if exists or product string of the device
for device description instead of the one from the .inf file
* Fix: Crash occurs if feature unit is specified explicitly via .inf file
* Fix: Linker optimization eliminated gDriverVersionString which we want to keep in the .sys
-----------------------------------------------------
V4.45.0 (June 18, 2018)
-----------------------------------------------------
* New: driver tested on Windows 1803 (RS4)
* Chg: volume input/output channel names now customizable in control panel
* Chg: ASIO device is now selectable, new API functions:
TUSBAUDIO_SetPreferredASIODevice, TUSBAUDIO_ClearPreferredASIODevice
* Fix: volume control for device with master only control was not working
* Fix: sound device name, visible via DirectKS now uses the string from the USB terminal/control interface

 

[Biblob]

* Fix: Spy tool doesn't work correctly with virtual channels
* Chg: improved Spy tool statistics

You lot are being spied on by the government *joke*

 

[la2ygoo]

The noise in Topping Dx3Pro is divided in two types of noise. One is from outside power supply, aka adapter (we talked about it last time)
and the other one is from inside unit power supply. I studied internal noise yesterday and I found at the RCA output a fundamental frequency of 570khz. After that I was looking at the input power supply votage (15Vdc) with the DAC "on" and I found the same fundamental frequency in the voltage ripple. Here, at the input, the voltage ripple has 120mVp-p ! So, the first voltage regulator from inside the unit has a big charging currents at 570khz. Here is the culprit. I looked to see what regulator it is and I found TPS54331. In the datasheet we found that: TPS54331 it is a step-down DC-DC converter with 570khz switching frequency. It has at the output an LC filter. This capacitor must to be large enough (in Dx3Pro is) and very important, to have low ESR at the frequency of interest. Unfortunately Topping put an electrolytic capacitors here. Until 100khz that type of capacitors could be very good but at 570khz and upper they lose their efficacy. Probably it was better few low-ESR tantalum capacitors in parallel or ceramic high value again in parallel mode. The best would be an linear regulator but the space is too small.

570khz，it make sence The noise con't be hear,and the audio analyzer con't measure it.

 

[Vosya]

You lot are being spied

Thesycon's "TUSBAudio Spy Tool" is a great thing. It is unfortunate that I could not find it for driver version 4.xx.

 

[777]

@amirm's analyzer has 1MHz bandwidth so it's enough as well.
on the contrary your osc is far from enough to measure something 24bits precise.

)))))))))))))))))))))))))))))))))))))))))))))))))))) You joke, ok ?