Review and Measurements of Topping DX3Pro DAC and Headphone Amp

[Stonetown]

Does anyone have an explanation for the paradox that headphone output is significantly better for SNR and THD?
Indeed, in the block diagram Topping headphone amplifier is after RCA. Therefore, THD and SNR can only be the same or worse.

View attachment 20711

amirm , can your jack3.5 cable be better than RCA?

As an old circuit designer my idea is that the noise we see at higher frequencies on the RCA-output is noise from cpu and other logical circuits reaching the Op-amps inputs capacitively via the PCB-layout or via the Op-amps supply or grounding pins. It is hard to get rid of this problem completely at the low levels we are measuring here, and it might be so that impedance levels are changed in different modes (DAC och Headphone or Pre-Amp) and therefore the amount of capacitive coupling to inputs is changed between the modes. It may also be that same noise is reaching the head phone Op-amp in somewhat antiphase and therefore the sum of the amplitude of the noise levels at the head phone output will be less. It can be quite a tricky work for the designer to improve such things further, but should not be impossible.

The noise can also sometimes originate from out of band high frequency noise which is translated to the audio band via intermodulation/mirroring on the testtone and nonlinearities in the following amplifiers, so not so easy sometimes to find where noise is coming from. But under these circumstances it could be improved by a following bandwidth limiting stage.

Just some speculative thoughts. Cheers!

 

[Veri]

Hi guys, another newbie question:

Which bit rate do you guys recommend for the DX3 Pro? I'm using it with a Beyerdynamic T1 and use it mainly for listening to 320kbps music (Streaming service) and playing games, lots of.

Currently i'm using 24/96 but the standard setting appears to be 32/44.1. Any recomendation about which bit rate to use?

Just use 32/44 and use exclusive mode of your video or media player for higher rates which will adapt the rate to whatever the source is.

Most games and music is 44Khz anyway, the DX3 filters/oversamples it to a higher rate so don't worry about that, 44 is perfect really..

 

[chrisliuboy]

Just use 32/44 and use exclusive mode of your video or media player for higher rates which will adapt the rate to whatever the source is.

Most games and music is 44Khz anyway, the DX3 filters/oversamples it to a higher rate so don't worry about that, 44 is perfect really..

In fact, one thing to keep in mind is don't use exclusive mode while playing games, once the media player switch to a different sample rate file like 96khz...It will mute everything else which was playing in 44khz like the game. Or you could always resample to 44khz in the media player first.

 

[chrisliuboy]

Hi guys, another newbie question:

Which bit rate do you guys recommend for the DX3 Pro? I'm using it with a Beyerdynamic T1 and use it mainly for listening to 320kbps music (Streaming service) and playing games, lots of.

Currently i'm using 24/96 but the standard setting appears to be 32/44.1. Any recomendation about which bit rate to use?

Only one thing is Dx3 makes the crackle noise of some voice recording in songs/video very obvious compare to my other audio source (S9+, old dac & on-board sound card), I have to use my X1 instead of the bright sounding cd900st with DX3 now. Since you are using the T1, love to know if you feel about the crackle noise.

 

[amirm]

@amirm , can your jack3.5 cable be better than RCA?

It is not the cable but I do wear out the ends due to very heavy use and sometimes they don't make solid connection and reduce measured performance.

 

[exaudio]

@yue , my DX3 was stuck in standby mode even when auto-power-on mode was turned off (A-C). I posted a short video to youtube to demonstrate. In the video I reset the unit so auto-power-on mode was reset to the default on (A-O). I think I said it wrong in the video, but I did try both modes. The video is here:

I eventually did get it to come out of standby by trying a different 15 VDC power adapter I had laying around. Unfortunately when it did power back on, the audio was severely distorted. I contacted Shenzhen Audio and they kindly offered to replace it so it is on it's way back to them.

 

[chrisliuboy]

@yue , my DX3 was stuck in standby mode even when auto-power-on mode was turned off (A-C). I posted a short video to youtube to demonstrate. In the video I reset the unit so auto-power-on mode was reset to the default on (A-O). I think I said it wrong in the video, but I did try both modes. The video is here:

I eventually did get it to come out of standby by trying a different 15 VDC power adapter I had laying around. Unfortunately when it did power back on, the audio was severely distorted. I contacted Shenzhen Audio and they kindly offered to replace it so it is on it's way back to them.

Sorry I am just new here, I am using another 15V 1.66a adpoter also, I don't know how distortrd sound is define. I used some 10db sub-bass boost someday, after stopping using it, I feel I could hear more crackle noise from bad recording, but from every headphone I could hear the crackle clearly then before(maybe just my memory is bad), is that counts?

 

[MattG]

It is possible it is caused bynbug in Topping's implementation of the auto-standby mode.
To isolate the problem. Is your device running at auto-standby mode? Could you please turn it off and see if the problem continues?

I read the manual in both English and Chinese and here is my understanding:

AO is to turn on “auto power off” mode and this is the factory default setting.

AC is to disable that feature. this is the setting you want to test.

I thought I had the auto power off disabled, but wouldn't swear to it. As of last night, though, it is set to "AC".

Minor side gripe: as per the discussion above, it's annoying that there is no way to query the setting, as pushing the "Auto" button appears to immediately switch it, without first showing the current setting.

Also, mine doesn't get "stuck" in this state: pushing the power button on the remote or the button on the unit itself immediately turns it back on.

I sent an email to Topping about this, still no reply.

Does anyone use the DX3Pro on linux? Whenever I stop/start playing something on Linux, there's a faint buzz sounds, and the USB sign flash (which means no signal I think?). How to deal with that?

I use mine connected via USB to a Raspberry Pi running DietPi Linux. No issues such as you describe. My only issue is randomly entering sleep/powersave mode, as discussed above. Fortunately it's (so far) infrequent enough to be only a minor annoyance.

 

[Stonetown]

@yue , my DX3 was stuck in standby mode even when auto-power-on mode was turned off (A-C). I posted a short video to youtube to demonstrate. In the video I reset the unit so auto-power-on mode was reset to the default on (A-O). I think I said it wrong in the video, but I did try both modes. The video is here:

I eventually did get it to come out of standby by trying a different 15 VDC power adapter I had laying around. Unfortunately when it did power back on, the audio was severely distorted. I contacted Shenzhen Audio and they kindly offered to replace it so it is on it's way back to them.

WARNING to all of you thinking of trying different DC-adapters:
Many older adapters are unregulated and the stamped voltage only applies under a certain load, so if the load is less the voltage will be much higher. Such 15 Volt adapter kan with less load go up to 20 Volts easily or more and destroy your unit.

To be sure you must measure the voltage from the adapter with no load, before using it.

BE CAREFUL!

Cheers! /Per

 

[Stein]

WARNING to all of you thinking of trying different DC-adapters:
Many older adapters are unregulated and the stamped voltage only applies under a certain load, so if the load is less the voltage will be much higher. Such 15 Volt adapter kan with less load go up to 20 Volts easily or more and destroy your unit.

To be sure you must measure the voltage from the adapter with no load, before using it.

BE CAREFUL!

Cheers! /Per

Would a LPS pass certification fix these issues? https://www.cui.com/blog/overview-of-limited-power-source-lps-requirements

 

[finneybear]

My DX3 is not delivered yet.
I myself am very interested in comparing DX3 sound in direct dsd + HQPlayer.
I am sure that the sound quality will be high. Because the main contribution to the sound will be from HQPayer.

Looking forward to seeing your comparison result.

I just ordered a DX3 Pro as well.

I did a comparison between DSC2 and LKS MH-004 (considered as one of the best implementations of ES9038 Pro) with DSD512, HQPlayer poly-sinc-xtr filter. DSC2 has better image with a warmer/smoother sound. I was using a special version of Amanero board which has separate power supplies for each clock.

My take is that those ESS chips are using current switch banks for the output. Being a CMOS device, it's hard to get matching switches hence the conversion errors. ESS uses dynamic element matching by swapping switch order to reduce errors. However, this may not work well for all of the scenarios. This is especially true for mid level intermodulation case where half of the switches are on. This is the worst case for DEM and may explain the showing up of ESS hump. Mobile ESS chips, those K2M, Q2M ones, have less number of switches hence DEM is less effective. For Pro grade chips, there are four output banks for each channel and ESS assigned the output banks based on measurement result to reduce the mismatches (they call this THD compensation). With correct bin sorting, a good Pro based DAC may get away from the hump.

On the other hand, DSC2, with 0.1% resistors and consistent NXP shift registers, it is simply more accurate. The output transformer helps a lot, too.

AKM chips use switching capacitors for the output. Switching capacitors have higher noise yet active capacitors are in general more consistent across the board. It also has less switching noises. This is why I have high hope for DX3 Pro on DSD512. Too bad that that board has too many digital noises.

 

[finneybear]

Does anyone have an explanation for the paradox that headphone output is significantly better for SNR and THD?
Indeed, in the block diagram Topping headphone amplifier is after RCA. Therefore, THD and SNR can only be the same or worse.



amirm , can your jack3.5 cable be better than RCA?

It's all about LPF. LPF is the hardest part to design in a DAC. There is no such thing as a perfect LPF. There will always be some HF noises leaking through. For an active LPF, if the HF has a lot of energies and the device does not have enough gain bandwidth, the linearity may get hurt.

OPA1612 is an excellent choice for LPF for its relatively high gain bandwidth. Unfortunately. they pack too many different digital functions into such a small board in DX3, digital noises will leak into the analog signal paths and pass through the LPF hence the distortions in harmony multiples.

The next stage, OPA2140 is a slower device so in some sense, it serves as a second LPF stage. Another good thing is it has JEFT inputs which would block HF noises from going into the OP. Then the final headphone output transistors will further smooth out signals.

I would say Topping engineers really know this thing and are smart enough on choosing the right parts.
For such a small PCB with so many functions in, they have done a very good job.

BTW, I do not feel its worthy to put in a low phase noise NDK clock except that you can figure out a way to add local power supply to it.

 

[Cidious]

I have 2 units running and no problems whatsoever.. it's a bit sad to read all this and also a bit scary..

 

[bellyofthebizd]

I have 2 units running and no problems whatsoever.. it's a bit sad to read all this and also a bit scary..

i ve got one for about an month with no problems exept sometimes it stucks in dsd mode when i close music player playing dsd file and there is no relay click, then after turning dac on it stays on dsd playback and there is no way to play anything until power it off with dc plug, but thats all, no big deal for me.

 

[finneybear]

doesn't necessary --- lks mh-004 looks just like audio-gd, with piled up components on a nice looking PCB, and gives no measurement at all. I bet it would measure very poorly, much poorer than DX3.

If you want the best implementation of ES9038Pro, Oppo Sonica DAC (which is based on the same circuitry as the BD-205 player) is the best you can find in the market.

Well, Sonica is one sensitive topic and I have to give out comments carefully as here some people may feel offended.

To make it simple, had Sonica then dumped it quickly. There's a big gap between Sonica and MH-DA004.
I am listening with DSD512 most of the time now. The USB interface of Sonica is just not that good.

I would call Sonica's DAC section a textbook like design. Decent parts with good ground plane separation, and good shielding and then... that's it. When you have a big PCB, you can afford all kinds of luxuries.

We do AP tests mainly to spot design problems. After no obvious design faults and the performance reaches certain level, we will focus on other aspects of the design. OP based output circuit is easier to do and you can get good performance numbers quickly. Discrete output stage can be very good, too, if you know how to do it right. With less NFB, it just sounds different.

Here is one example - It's a PCM1794 based DAC we designed 10 year ago for fun. Remember that 10 years ago we did not have ultra low noise voltage regulators, and the performance was bound by the limitation of PCM1794.

The output is full discrete based. As you can see, other than the harmonics, the output is very clean. We did find out the source for the 3rd harmony spike, and eliminated it later. We do like a bit higher 2nd harmony though. Again, this is a 10 years old design. Still sounds very good today.

If you look at Oppo's FFT, though the overall THD is very low, those spikes around 1KHz are a concern to me. Potentially this will mean the OPs are under strains. Who knows what will happen when you have true music, multiple frequencies come in at the same time?

When you run ES9038 Pro in full mono mode, the output current is around 100mA, and this is too much for most OPs.
This is why discrete I/V becomes a serious choice. To do it right, good device matching is a must, and the cost can go up quickly.
This is also why the LKS is double of the price of Gustard.

I got to know audio-gd when they just started the business. They were doing discrete OP OEMs for Burson back then.
It's basically a workshop selling parts, lots of parts. Other than that, I do not want to talk too much about them. LOL

DSD512 does have it magic. Once you go DSD512, you may never want to come down again. Even DSD256 is pretty good.
AKM chips have native DSD support and they do it in a way different from ESSes. I plan to feed the output of DX3 to a pair of output transformers when running DSD512. I have the gut feeling that this thing can be a giant killer.

 

[Stonetown]

Would a LPS pass certification fix these issues? https://www.cui.com/blog/overview-of-limited-power-source-lps-requirements

Probably... but to be sure and not jeopardizing your DAC just hook up a DC-voltmeter at the DC-plug before using it, and check that the voltage is same as the original supply within lets say 0,2 Volts compared to the original. Also check the polarity, because sometimes plus and minus are reversed, and you dont want any smoke... Thats what I would do and I am an electronic engineer, and have learned about Murphy's law (https://en.wikipedia.org/wiki/Murphy's_law) so never trust Verification is always a god bet! If you cannot check/verify, stay with the original supply!
Cheers, Per

 

[Pavel]

I plan to feed the output of DX3 to a pair of output transformers when running DSD512. I have the gut feeling that this thing can be a giant killer.

Yes, I also think it can make a very good sound. My colleague compared the AK4490 direct DSD256 and DSC2 DSD256. The output in both variants was the same transformer. And the sound was similar in quality. Therefore, I look forward to DX3 for checking on DSD512 and transformer output.

 

[finneybear]

Yes, I also think it can make a very good sound. My colleague compared the AK4490 direct DSD256 and DSC2 DSD256. The output in both variants was the same transformer. And the sound was similar in quality. Therefore, I look forward to DX3 for checking on DSD512 and transformer output.

DSC2 has the advantage with 64 levels when 4493 has capacitor banks. 4493 may also have less switching noises. Will see, which one sounds better.

In the meantime I am thinking about improving DSC2 by adding a tiny decoupling cap at the flop output to reduce the switching spike. Probably add a local power supply to XO, too.

 

[splkn]

Yes, I also think it can make a very good sound. My colleague compared the AK4490 direct DSD256 and DSC2 DSD256. The output in both variants was the same transformer. And the sound was similar in quality. Therefore, I look forward to DX3 for checking on DSD512 and transformer output.

Pavel, what transformers do you plan to use?

 

[MattG]

could you run dmesg on your linux and see if there are signs of USB disconnect-reconnects. probably you won't see that as when it's in standby mode the usb interface is still working and shows as a valid usb device in OS. but worth a try anyway.

Ahh, great "why didn't I think of that" idea!

Unfortunately, I don't see anything telling. Only three instances of USB-related messages. The first is when the system booted and initially discovered the DX3 Pro. The next two are exactly when I swapped out the power supply (one to use the alternate supply, and the other when I put the stock power supply back). In other words, only seeing exactly what I would expect.