• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Topping DX3Pro DAC and Headphone Amp

Klenfo

Member
Joined
Dec 1, 2018
Messages
43
Likes
14
When I am hearing about relay clicking with browser and such, I am assuming that you are not using WASAPI or ASIO since they don't support either interface
Yes with DS: Primary Sound Driver in foobar2000 and in browsers relay cliсks also present, problem not in ASIO or WASAPI.
I hope there will be a new firmware with fix.
 

Wolven

Active Member
Joined
Nov 10, 2018
Messages
226
Likes
113
You guys are confusing me now. :)

Only dedicated music players like foobar, roon, etc. can use Wasapi. When I am hearing about relay clicking with browser and such, I am assuming that you are not using WASAPI or ASIO since they don't support either interface. If you are using WASAPI with exclusive mode, it would not allow any other app to attach to the dac and hence cause it to do anything.

So what is exactly is the situation?

There are two problems:
A. First one is small. Hit pause in Foobar - click. When playing music, I can go on the internet at the same and of course no problem. Music continues to play all is good.
B. Stoop music in Foobar. Now, go on internet. Everytime, you load a page, click, change page click. Go to youtube turn on a video, click, stop video click. start the same video again - click

I probably should have said the problem is the usb driver of dx3pro, and how it interacts with ASIO/WASAPI.
I'm no specialist
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,395
Location
Seattle Area
There are two problems:
A. First one is small. Hit pause in Foobar - click. When playing music, I can go on the internet at the same and of course no problem. Music continues to play all is good.
B. Stoop music in Foobar. Now, go on internet. Everytime, you load a page, click, change page click. Go to youtube turn on a video, click, stop video click. start the same video again - click

I probably should have said the problem is the usb driver of dx3pro, and how it interacts with ASIO/WASAPI.
I'm no specialist
So this happens with WASAPI/ASIO plug-in in foobar or without?
 

Klenfo

Member
Joined
Dec 1, 2018
Messages
43
Likes
14
There are two problems:
A. First one is small. Hit pause in Foobar - click. When playing music, I can go on the internet at the same and of course no problem. Music continues to play all is good.
B. Stoop music in Foobar. Now, go on internet. Everytime, you load a page, click, change page click. Go to youtube turn on a video, click, stop video click. start the same video again - click
Exactly. Looks like device goes into some standby mode after the signal disappears - relay click. Then start sound - device returns to ON mode - relay click again
 

TP-Wong

Member
Audio Company
Joined
Dec 3, 2018
Messages
44
Likes
242
Location
Guangzhou, China
Hello every one. I'm Hailin Huang, sales manager of TOPPING. I have engineering background for many years so I think I may be the suitable person in our company to communicate directly with you.

Sorry for our late reply, because our response comes too late, that makes every one here feel confused and also make Amir questioned.

I would like to introduce the problem of "click" here. And if any one have other questions, you can ask me here.

What makes the click? Is it a defect unit?
AK4493 will output a pop noise when the sampling rate changes, especially in DSD. To void this noise, we test many method including semiconductor switch and relay, at last we decide to add a relay because it did better in avoiding the pop noise. But the relay also bring the click noise. You can hear the click from the DX3 Pro itself, and also a little from the headphone(If the sensitivity of headphone is high). Because this is our setting, so every DX3 Pro has click sound, the unit with click sound is not a defect unit.

When it will click?
The relay is controlled by both XU208 and AK4118. When there's a audio stream, the relay closed, and if there's not a audio stream, the relay cut the output. So if you continue listening to music, you will not hear the relay so often. But if you doing normal operation to your PC withoud a continuous audio stream, the relay will works very often.

But when the sampling rate did not change, the relay also keep working.
Yes, that's because we are not able to judge the sampling rate is same or change when the stream comes, so we cut every time when the stream over and on.

Is there a solution?
I think yes. We will try to find a solution from XMOS. If it did not send out "cut off" signal when sampling rate did not change, this will be able to reduce most of the click sound. We already ask the XMOS firmware supplier try to solve this problem, this will take serval days and we will tell progress and results if necessary. I think the best solution will be a firmware update for DX3 Pro, we will keep working on this.
 

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
I get that you are frustrated but you aiming that at me isn't going to resolve the issue. You need to wait until Topping responds and then we know where we stand.

And yes, it is my site and I will continue to make recommendations on products that I find highly performant.

Clicking noise is not your fault, you didn't face it with your unit. You can't be blamed for this. You can't be blamed for anything - as you mentioned this is YOUR site (though it sounded a bit desperate), you don't force anyone to buy something or do anything, this is just great testing site and targeting/blaming the host is very stupid IMHO.

With all due respect to you and all your great job, only thing that irritated me was not putting 15,5 bit linearity test and poor toslink performance in the review. That would save me some money on shipping back the unit, because if I had seen this tests earlier I wouldn't have bought it. You can have full image of performance only when you go through the whole thread and finding it out instead of just having a look at the review itself. In my opinion this makes the review itself a bit incomplete or incoherent.

When yggdrasil achieved 16 bit linearity in one of your reviews, you commented it like this: "Same theme emerges here with the Gen 2 board not having as wild of variation/error as Gen 2. That said, this is unacceptable resolution and fidelity for a DAC. [...] If you play high-resolution audio, this DAC is not for you."

Reference link:
https://www.audiosciencereview.com/...ents-and-review-of-schiit-yggdrasil-dac.2358/

So it is hard to call this topping "highly performant" to listen only 16 bit files that it can barely play properly and make use of 16 bits. For me it's not a big deal beacuse 90% of my collection are 16 bit 44,1k files, but this is advertised as high-resolution audio dac. Eventually I will send this back and buy something that I know it sounds great by my ears. Will only check this out as extra DAC for future comparisons.
 

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
Hello every one. I'm Hailin Huang, sales manager of TOPPING. I have engineering background for many years so I think I may be the suitable person in our company to communicate directly with you.
But when the sampling rate did not change, the relay also keep working.
Yes, that's because we are not able to judge the sampling rate is same or change when the stream comes, so we cut every time when the stream over and on.

This is definately wrong implementation and thus all problems.

Is there a solution?
I think yes. We will try to find a solution from XMOS. If it did not send out "cut off" signal when sampling rate did not change, this will be able to reduce most of the click sound. We already ask the XMOS firmware supplier try to solve this problem, this will take serval days and we will tell progress and results if necessary. I think the best solution will be a firmware update for DX3 Pro, we will keep working on this.

It should be done like this from the beginning. All dacs work like that - I tested like 5 different dacs with 4490, 4495, neither of them had this issue.

Keep up the good work and prepare firmware upgrade for us ! :)
 
Last edited:

yue

Active Member
Joined
Jul 21, 2018
Messages
275
Likes
294
Is there a solution?
I think yes. We will try to find a solution from XMOS. If it did not send out "cut off" signal when sampling rate did not change, this will be able to reduce most of the click sound. We already ask the XMOS firmware supplier try to solve this problem, this will take serval days and we will tell progress and results if necessary. I think the best solution will be a firmware update for DX3 Pro, we will keep working on this.

Well again this further implies that Topping needs to control firmware development within your own hand, as many other Chinese vendors do. This will certainly help to avoid most issues like this before releasing the product and expedite the fixing process when user run into such issues.
 
Last edited:

Wolven

Active Member
Joined
Nov 10, 2018
Messages
226
Likes
113
Clicking noise is not your fault, you didn't face it with your unit. You can't be blamed for this. You can't be blamed for anything - as you mentioned this is YOUR site (though it sounded a bit desperate), you don't force anyone to buy something or do anything, this is just great testing site and targeting/blaming the host is very stupid IMHO.

With all due respect to you and all your great job, only thing that irritated me was not putting 15,5 bit linearity test and poor toslink performance in the review. That would save me some money on shipping back the unit, because if I had seen this tests earlier I wouldn't have bought it. You can have full image of performance only when you go through the whole thread and finding it out instead of just having a look at the review itself. In my opinion this makes the review itself a bit incomplete or incoherent.

When yggdrasil achieved 16 bit linearity in one of your reviews, you commented it like this: "Same theme emerges here with the Gen 2 board not having as wild of variation/error as Gen 2. That said, this is unacceptable resolution and fidelity for a DAC. [...] If you play high-resolution audio, this DAC is not for you."

Reference link:
https://www.audiosciencereview.com/...ents-and-review-of-schiit-yggdrasil-dac.2358/

So it is hard to call this topping "highly performant" to listen only 16 bit files that it can barely play properly and make use of 16 bits. For me it's not a big deal beacuse 90% of my collection are 16 bit 44,1k files, but this is advertised as high-resolution audio dac. Eventually I will send this back and buy something that I know it sounds great by my ears. Will only check this out as extra DAC for future comparisons.

So, how bad of a problem are we talking here with this linearity?
I have about 100 albums in high resolution, most of those 24/96 some 24/192.
Is this DAC useless for that?
 

yue

Active Member
Joined
Jul 21, 2018
Messages
275
Likes
294
This definately wrong implementation and thus all problems.



It should be done like this from the beginning. All dacs work like that - I tested like 5 different dacs with 4490, 4495, neither of them had this issue.

Keep up the good work and prepare firmware upgrade for us ! :)

Topping outsources their firmware to a third party supplier, and in my humble opinion the supplier is not very reliable. The supplier messes various firmware configuration up, and is reluctant to fix them. They cause quite a few issues in OSes or music players. If I were topping I would hire my own firmware developer and clean up all the bugs.
 

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
So, how bad of a problem are we talking here with this linearity?
I have about 100 albums in high resolution, most of those 24/96 some 24/192.
Is this DAC useless for that?
Correct me someone if I am wrong, but you will be hearing some noise instead of extra 8 bits of information.

But for 16 bits files it should be audibly transparent as amir posted earlier in this thread.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,395
Location
Seattle Area
When yggdrasil achieved 16 bit linearity in one of your reviews, you commented it like this: "Same theme emerges here with the Gen 2 board not having as wild of variation/error as Gen 2. That said, this is unacceptable resolution and fidelity for a DAC. [...] If you play high-resolution audio, this DAC is not for you."

Reference link:
https://www.audiosciencereview.com/...ents-and-review-of-schiit-yggdrasil-dac.2358/

So it is hard to call this topping "highly performant" to listen only 16 bit files that it can barely play properly and make use of 16 bits.
I explained the issues with linearity very clearly in the review. I mentioned that it is indicating increase in noise, not error. Here is a composite of Yggdrasil and DX3 Pro:

1543823855770.png


The DX3 Pro is far more benign than Yggdrasil. You are going to get noise at extreme low levels anyway. So while I prefer to not see the rise in noise in DX3 Pro, it simply is not the same problem as noted in Yggdrasil.

I also explained earlier that I am still investigating this as I have not see the response like I am seeing in DX3 Pro before. So it might be an instrumentation issue.
 

Klenfo

Member
Joined
Dec 1, 2018
Messages
43
Likes
14
When there's a audio stream, the relay closed, and if there's not a audio stream, the relay cut the output. So if you continue listening to music, you will not hear the relay so often. But if you doing normal operation to your PC withoud a continuous audio stream, the relay will works very often.
Why to cut the output after stop sound? For example in youtube it is constantly play-stop = click-click. No changing sample rate over there, I think it is no need to off relay. Please fix it
 

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
I also explained earlier that I am still investigating this as I have not see the response like I am seeing in DX3 Pro before. So it might be an instrumentation issue.

I think that you should measure it again using production unit because it could have changed only for better. Maybe there was a problem with engineering sample. Maybe some measurements error. Hard to believe in 15,5 linearity with such nicely performing unit in other areas.

Anyone has it yet in US and can send it to amir for remeasuring ??
 
Last edited:

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
Why to cut the output after stop sound? For example in youtube it is constantly play-stop = click-click. No changing sample rate over there, I think it is no need to off relay. Please fix it
Yep, in windows you also have a lot of different asynchronous sound events like communicators etc which can generate random sounds at a time. This can be turned into clicking hell. That's why it shouldn't be implemented like this, but I nailed it in this thread some time ago as one of possible reasons regarding not comparing what was played before what many other dacs do and that are silent as long as the new song (sound in general) has the same bit depth + sampling frequency.
 

KT88

Member
Joined
Dec 2, 2018
Messages
13
Likes
11
bina998, this 15.5b is THD+N, where in this case its the N that is dominant, not the THD. Also, this takes into account BW of 90KHz, and not just the 20KHz you see in the figure, most of which is not audible. Would be interesting to see if there is something funny going on outside the 20KHz limit which might affect this measurement.
 

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
Why to cut the output after stop sound? For example in youtube it is constantly play-stop = click-click. No changing sample rate over there, I think it is no need to off relay. Please fix it
Because if you can't tell what is next format you could hear loud pop/scratch when it's actually different than before. So without comparing it first or some other mechanism implemented, they decided to protect it every single time the new song starts to play to avoid this annoying sounds.
 

bona998

Member
Joined
May 20, 2018
Messages
44
Likes
8
bina998, this 15.5b is THD+N, where in this case its the N that is dominant, not the THD. Also, this takes into account BW of 90KHz, and not just the 20KHz you see in the figure, most of which is not audible. Would be interesting to see if there is something funny going on outside the 20KHz limit which might affect this measurement.
I didn't analyze the plots. I was just reffering to the amir's post when he claimed this unit has 15,5b linearity as I was digging through the thread.

You can ask amirm to have it measured once he's in possession of production unit (this outside 20kHz).

btw it's not "bina998"
 
Last edited:

Wolven

Active Member
Joined
Nov 10, 2018
Messages
226
Likes
113
Because if you can't tell what is next format you could hear loud pop/scratch when it's actually different than before. So without comparing it first or some other mechanism implemented, they decided to protect it every single time the new song starts to play to avoid this annoying sounds.

My Audioengine D1 was like that with WASAPI, the screeching noise was a nightmare. Good thing, I wasn't blasting music or my speakers probably would have been blown.
 

Klenfo

Member
Joined
Dec 1, 2018
Messages
43
Likes
14
Because if you can't tell what is next format you could hear loud pop/scratch when it's actually different than before. So without comparing it first or some other mechanism implemented, they decided to protect it every single time the new song starts to play to avoid this annoying sounds.
In windows format is set in Device Properties -> Advanced -> Default Format
So it's not need to compare - not need relay click
 
Top Bottom