Review and Measurements of Topping DX3Pro DAC and Headphone Amp

[Krunok]

Well, the way I see it the best way to learn is to teach

The way I see it you first have to learn and only then you may teach, assuming that besides the knowledge you will be teaching you also have a skill to transfer your knowledge to your students.

And you should never stop learning, of course.

 

[mikehoopes]

“...Come here to have fun, be ready to be teased and not take online life too seriously.” That’s the other foundational concept here.

And yes, I come here for the measurements. All this talk of improving the bass response, “audio grade” components, with theories of operation (thank you, @finneybear), and subjective reports, is not unwelcome, but I personally would like to see a through-line of “Ok, I hear a change, now let’s devise a measurement that strongly correlates the subjective observation with the change. While we’re at it, let’s also compare it to known thresholds of audibiilty.” That better advances the art.

 

[Cidious]

It's so tiring to constantly read here that measurements are considered to be 1 on 1 connected to a pleasant listening experience... some of the older vintage gear measures not nearly as good as the modern day stuff but definitely sounds better... At least to my ears... And many others with me. I wouldn't say this is always the case but it's just an example.

Throwing everything on measurement is just nuts just like throwing everything on one pair of ears is useless... there is a way in between.

Some audible differences are hard to be measures. Just like some measured differences are barely audible...

Measurements are great for analysing and troubleshooting a design. But most of these measurements don't translate to a direct audible conclusion.

I've said it before and I'll say it again. I would like to ask the measurement police here to explain to me how those measurements translate into audible changes. How these changes can be described. Then suddenly these measurements will be much more relevant. But so far I've not gotten any response to the question and nobody really seems to know how these measurements translate into listening experience. But measurements are the holy grail... Everything else is irrelevant.. right ?

 

[trl]

[...]

Did also see the Burson playmate and swing that look interesting swing for me as it has a few more connections

It has a bit lower background noise than the latest PLAY board revision, which is great. They moved from TO-252 to SO-8 chip regulators, still boost-converters. The negative regulator is from TI and it's less noisy than the XLSEMI one (in the past, the positive regulator was having a lower noise than the negative one).

It really seems a complete device, with several digital filters. The noise with NE5532 or LME49720 opamps is very low; with my 103dB/mW 16Ohms IEMs I can hardly hear any noise...and only if inside the room there are no ambient noises at all.

I can post some pics if you want...but for it's price and for a 10V RMS output voltage and 2W/ch. I think it is a good combo, better than PLAY for sure. There's a relay to switch from headphones to RCA and an internal gain switch (a digital one, so no background noise is adjusted because is not analogue gain).

 

[Killingbeans]

It's so tiring to constantly read here that measurements are considered to be 1 on 1 connected to a pleasant listening experience...

I don't think I've seen a single person on this forum claim that. Most people here seem to agree that measurements are considered to be 1 on 1 connected to audible or non-audible differences and to good or bad/lazy electrical engineering (depending on the claims of the designer). Whether those differences make the sound pleasant or unpleasant is a completely separate discussion.

I would like to ask the measurement police here to explain to me how those measurements translate into audible changes. How these changes can be described. Then suddenly these measurements will be much more relevant.

The mods being displayed i this thread are all aimed at lowering the noise floor and minimizing distortion. If measurements of the modified DAC shows zero change in noise or distortion, can we then agree that these mods have been done in vain? If the only purpose of modification is to make audible changes then why not just solder random components between random solder joints and see what happens?

 

[amirm]

It's so tiring to constantly read here that measurements are considered to be 1 on 1 connected to a pleasant listening experience...

That is never been said. Measurements do however show 1:1 correlation as to whether the sound has changed. To the extent measurements show no difference whatsoever, then whatever change has been made is not material to the sound. It is an impossibility that sound has changed in an obvious manner but the measurements show no change.

 

[finneybear]

Very interesting. Could you draw the schematic ?

Here is for your reference:

 

[finneybear]

And you can tell all that without measurement, only by ear?

P.S. I'm sure changing stock power supply would improve the sound even more. Just be sure to use directional power cable which has been properly burned-in.

So you are listening to music with measurement instruments?

Since you believe measurements, can you please tell me why stock DX3 is measured so well yet the sound is not very good hence some owners have started using DX3 as a paperweight? Does that mean the particular owner's ears are not as good as measurement instrument?

Very surprised to read your comment on the power supply... again, since you like to read measurement result... have you read into Amir's test report? He could not see much difference after switching to a lab power supply?

The reason is very simple, based on the part use and design in stock DX3, the problem is not in the external power supply. This is why the measurements cant see much difference. You have to fix problems in DX3 before the benefit of better external power supply starts to show up.

My mod was originally aimed to resolve the stability issue of DX3 and to fix some questionable design decisions. In other words, I am trying to put better engineering into the machine.

Is good engineering a problem to you? Doesnt good engineering go before good measurement?

 

[finneybear]

Sorry if this has already been asked and answered... Can you send the modified unit to @amirm so he can quantify the differences vs stock units?

I have been thinking about it, too. I do have access to AP. My company should have a lab or two having the latest AP, too. Still, since everybody sets up the AP slightly differently, the skill of the operator matters, too, it's better to have Amir do the test to ensure a consistent result.

Just did an A-B test with ES9038 Pro based LKS MH-004, got some very interesting result. I have decided to swap the XOs in DX3 to see whether I could improve it further more. Yeah, DX3 is now sooooo good.

 

[finneybear]

I can only hear it when set to headphone out. If I set it to line out or dac mode the noise goes away. Also, if I unplug a single RCA cord it goes away, it needs both plugged in to make the noise. Wouldn't EMI or ground noise still be audible if only one RCA cord was plugged in?

Still not sure if it's normal as it only occurs in the right ear piece and when set to high gain.

Yeah, that's what I had figured. The relays only disconnect the signal paths. not the ground paths so the noise will still come in from the ground line.

 

[finneybear]

You should let amirm measure your modified version of dx3pro

Actually I can already foresee what the test result will be. The THD will be a bit higher due to the different OP. The FFT will have far less noise spikes.

A lot of people are reading measurements in the wrong way. The measurements are mainly for helping engineers solving design problems. It takes some skills and design experiences to decipher the numbers and charts correctly. It's up to engineers to decide what they want and what they do not want. Really. it is not just how low THD is and what harmonics look like.

 

[finneybear]

Since you're fixing things with such ease, I'd recommend you to apply for a job on Topping engineers team.

It would be cool this enlightened wisdom designing products for us customers.

Dont blame on Topping engineers. They are limited by the part budget and the extra gimmicks the marketing requested to make the product sell.

$200 is the sweet spot on the market now. This will limit the part cost to around $100. The twin 4493 is really useless here but hey, it's a good sale point. Twin 4493 is required to compete against twin ES9038Q2M based DAC.

To me, I am free to upgrade the parts and do fixes which make more engineering sense. That's it.
Based on the parts I used, the DX3 retail price may run up to the $500 range. The market will be much smaller at that market segment.

Through forums, etc, I had got to know quite a few folks who are chief designers in China audio gear industry now. I used to help ESS doing chip designs, etc. Audio circuit is just a hobby to me. Be honest, there's not much money to make there. Dont do much on DAC these days actually. My main interest is rebuilding reel-to-reel and cassette decks.

 

[finneybear]

Thanks for the excellent report of the tweak. Using only 2 of the dac output sans output capacitor is an excellent idea. So you have to replace all of the existing parts around the opamp for the new filter values?
Also, my understanding is its a good idea to keep high frequency out of opamp input (which is one reason why opamp for dac is not so good). So I am wondering if replacing that zero ohm with a choke would be a good idea.
You might want to experiment with the decoupling cap for the opamp as well. Sometimes I found improvement in removing the ceramic cap and use a good ecap.

Yes, 4493 as voltage output DAC, paralleling them will not have much benefit. It will be better off to just use one set of outputs instead and to remove coupling caps. Now I have tons of strong and full bass! DX3 now can go very very low!

I did ready small caps to bleed HF switching noises right after the 750/249 ohm resistors. Yet after listening tests, I decided not to install them. It will be tricky to pull out ground lines to connect those caps, too. Probably just introduce more noises into the circuit due to the exposed ground line.

Yes, the smaller ceramic caps for the OP power are no good. After battling against ES9038 Pro, I feel I need to solve the jitter problem first.

 

[finneybear]

In that case you must to read only the first Amirm post with measurements, all the rest is blablabla . If you want the real science, well, you will find a lot of "life" outside of the Amirm standard measurements. We talking about that, not about snake oil. We talk about better caps, smaller noise of the power supply, proper impementation of the op-amps, etc,etc...real science. If you like only black and white science, without grey tones, it's ok, is your choice.

Just want to remind people, measuring an audio device can be a tremendous effort. There are so many things to measure. If you only look at the 1KHz FFT chart to dictate a machine's performance, most likely you will be led to the wrong conclusion. 100Hz, 400Hz, etc, are critical, too. Have you seen any 400Hz FFT here?

 

[finneybear]

“...Come here to have fun, be ready to be teased and not take online life too seriously.” That’s the other foundational concept here.

And yes, I come here for the measurements. All this talk of improving the bass response, “audio grade” components, with theories of operation (thank you, @finneybear), and subjective reports, is not unwelcome, but I personally would like to see a through-line of “Ok, I hear a change, now let’s devise a measurement that strongly correlates the subjective observation with the change. While we’re at it, let’s also compare it to known thresholds of audibiilty.” That better advances the art.

Actually the measurement is already done, as Pavel did. The 1KHz FFT will have far less noise spikes. Unfortunately the 1KHz FFT will not tell you much about bass performance.

The overall THD will be a bit higher due to a different OP. In a lot of cases, as a decent engineer, you can calculate ahead roughly what the measurement result will be.

 

[finneybear]

The mods being displayed i this thread are all aimed at lowering the noise floor and minimizing distortion. If measurements of the modified DAC shows zero change in noise or distortion, can we then agree that these mods have been done in vain? If the only purpose of modification is to make audible changes then why not just solder random components between random solder joints and see what happens?

The mod is mainly to fix some poor engineering decisions in DX3, not to lower noise floor or to reduce distortion, etc.

Another problem is the measurements done here, to have the result consistent across all machines, are limited to a few common tests - 1KHz FFT, etc. they give you a good indicator on how a DAC performs in general yet it will not tell you much about how a DAC perform at 400Hz, etc, for instance.

 

[Killingbeans]

So you are listening to music with measurement instruments?

Doesnt good engineering go before good measurement?

Isn't good engineering impossible without measurements?

A lot of people are reading measurements in the wrong way.

Totally agree.

The measurements are mainly for helping engineers solving design problems.

Absolutely not. Measurements are mainly for minimizing the amount of time and money being wasted, but they are also essential for solving design problems. An engineer who implements things without confirming their validity through measurements is sure to get fired (or at least deserves to be IMO).

 

[Killingbeans]

The mod is mainly to fix some poor engineering decisions in DX3, not to lower noise floor or to reduce distortion, etc.

Then what are the consequences of these poor engineering decisions? You claim that there's an obvious audible difference, meaning it will also influence some very obvious parameters?

 

[finneybear]

Isn't good engineering impossible without measurements?

I often like to compare this to bridge designs. Do you design a bridge, do a calculation first to see how much weight it can take/withstand test, or design a bridge first then test to see whether it will crash with cars passing? Surely a good engineering is first to ensure your design will have a good test result.

Absolutely not. Measurements are mainly for minimizing the amount of time and money being wasted, but they are also essential for solving design problems. An engineer who implements things without confirming their validity through measurements is sure to get fired (or at least deserves to be IMO).

Unfortunately, this is not necessary true, especially for consumer products, non-major brand ones.

Cant you see all of the stability problems people reporting on DX3? And I dont want to repeat this again and again. AP tests are not everything, and even the AP tests most people did are very limited. I have been using AP on and off since it's first generation.

 

[finneybear]

Then what are the consequences of these poor engineering decisions? You claim that there's an obvious audible difference, meaning it will also influence some very obvious parameters?

What if the audible difference needs 400Hz FFT to prove?