Topping DX5 Review (DAC & HP Amp)

[Bow_Wazoo]

I got 1.05.
But it's still doesn't work...

Audio - Google Drive

drive.google.com

 

[czAudiophile]

yep, thank you. I have Dx5 second revision. SO this works for me. Will try.
PS What website are using for importing headphones eqs?

 

[Jimster480]

yep, thank you. I have Dx5 second revision. SO this works for me. Will try.
PS What website are using for importing headphones eqs?

If you have the DX5 Gen II then that is a different product than this thread is dedicated for. As this thread is for Generation 1.
As far as a headphone EQ database.... I am not sure there is one but someone else hopefully can be more useful.

 

[staticV3]

What website are using for importing headphones eqs?

https://github.com/jaakkopasanen/AutoEq/tree/master/results
Contains PEQ presets in .txt format which you may be able to input directly into Topping Tune

https://www.reddit.com/r/oratory1990/wiki/index/list_of_presets/
Contains PEQ presets that you'd have to type into Topping Tune by hand, or first convert into .txt

https://autoeq.app/
Can be used to create PEQ presets in .txt format, with customizable parameters

https://squig.link/
and the 20+ variations of it (Burger menu->"Super* Review") can be used to create PEQ presets in .txt format, with customizable parameters.

 

[czAudiophile]

wow thank you so much.

 

[ICIETDIYEUR]

Hi everyone

Concerning the TOPPING DX5 (not Lite or II),

Has anyone carried out new measurements with the latest driver (v5.72.0 - from November 13, 2024) and the latest firmware (v1.42 - from July 6, 2023) ?

This should make it possible to see that the 'ESS HUMP' has disappeared, that the details of the JITTER have been adjusted and perhaps that the SINAD itself could have changed...

... In my humble opinion this DAC seems 'better' (I know it's subjective) than my old TOPPING E30, TOPPING E50, SMSL SU-1, SMSL PS200, SMSL D6s (same SINAD = 122) and even the TOPPING E70 VELVET (better SINAD = 123).

To remain as objective as possible, I tested the SMSL D6s and the TOPPING DX5 in comparative listening with a 1kHz signal in order to obtain by adjustment strictly the same output voltage on the two devices and under the same conditions then I listened several times to the same two titles that I know perfectly:

- Eric CLAPTON - Unplugged (DSD 128) 'Tears in heaven'
-> the sound of the triangle at 15 seconds is more natural with the TOPPING DX5.

- DAFT PUNK - Random access memory (studio master edition - 24 bits / 88.2 kHz) 'Get lucky'
-> in general, the 'definition' of the treble is much better with the TOPPING DX5 (among others).

Okay, it's probably still a 'psychoacoustic' effect but very sincerely I don't think so: can it be possible that there are indeed differences in your opinion ?

NB:

- TOPPING DX5 with Filter 1 (mode 1)
- SMSL D6s with Filter 4 (FL 4)
PS:

I noticed absolutely no 'heating problems' with the TOPPING DX5 with my infrared thermometer regardless of the location of the case.

In short, I find this device very good, especially at the price I got it NEW (€200)

 

[ICIETDIYEUR]

People hi

Amirm: have you re-measured the TOPPING DX5 with the latest firmware (v1.42)?

Normally the 'IMD/ESS HUMP' has disappeared and so have the TOSLINK jitter 'worries'.

I think it could be interesting to superimpose the different measurement tables of the TOPPING DX5 with the very latest TOPPING DX5 II.

We could very clearly compare the distortion in the higher frequencies between these two devices among others and thus see if the evolution proposed by TOPPING only concerns the PEQ/display (yes, the 12V trigger too) but not to the detriment of the SQ...

 

[staticV3]

People hi

Amirm: have you re-measured the TOPPING DX5 with the latest firmware (v1.42)?

Normally the 'IMD/ESS HUMP' has disappeared and so have the TOSLINK jitter 'worries'.

I think it could be interesting to superimpose the different measurement tables of the TOPPING DX5 with the very latest TOPPING DX5 II.

We could very clearly compare the distortion in the higher frequencies between these two devices among others and thus see if the evolution proposed by TOPPING only concerns the PEQ/display (yes, the 12V trigger too) but not to the detriment of the SQ...

Here's the updated IMD performance:

Source

The hump isn't entirely gone, but you can mask it at the expense of higher noise floor.

Apparently there are three options so you can pick your poison.

Remember also that Amir's THD+N vs Freq plots are measured at 90kHz BW:


Restrict BW to human hearing and the slight rise above 5kHz will be gone completely.

It never was a detriment to SQ anyway but 20kHz BW would make it even less so.

Now I'd suggest you lay your agenda to rest. It's plain irrational.

 

[ICIETDIYEUR]

Hi staticV3.

What projects?

What is irrational?

I'm being objective in comparing the results of precise measurements as we usually do, right?

 

[simmconn]

Hi staticV3.

What projects?

What is irrational?

I'm being objective in comparing the results of precise measurements as we usually do, right?

Agreed. Making claims without backing them up with data. Not sure who has an irrational agenda, lol:

Restrict BW to human hearing and the slight rise above 5kHz will be gone completely.

 

[jkim]

Agreed. Making claims without backing them up with data. Not sure who has an irrational agenda, lol:

I guess you misread the context in which @staticV3 was irritated by the post of @ICIETDIYEUR. To be honest, I am, too.

Keeps posting similar, pointless content, the primary purpose of which is just to have the ASR community endorse his purchase of a particular product, and more critically, his argument that the product he chose is better than others he did not. It was the Audient iD14 MKII before. Now it is the Topping DX5.

And @staticV3 was not making claims without objective grounds. He can make that statement because he knows how the THD+N versus frequency measurements are done and knows how to read the data beyond that particular context. The bandwidth (BW) of THD+N measurements has a significant effect. Because of noise shaping, the main purpose of which is to shift quantization noise from human hearing to ultrasonic range, almost all DACs' noise and distortion performance becomes substantially worse when the measurements BW is extended beyond the human hearing audio band. Also, THD+N results for fundamental tones b/w 5k and 20 kHz reflect distortion products beyond human hearing. That is why @staticV3 stated:

Restrict BW to human hearing and the slight rise above 5kHz will be gone completely.

And that is why we want to take into careful consideration its BW setting and implications of fundamental tone frequencies when interpreting a THD+N versus frequency plot.

In addition, some of us, including @staticV3, have compared countless AP measurements between Amir's at ASR and Wolf's at L7 Audio. In most cases, the noise level in Wolf's results is considerably lower and hence THD+N results are also better because he used his AP's performance mode. So, when comparing two devices b/w ASR and L7 Audio, it is possible that a difference is just due to a difference in measurement conditions, not because one device necessarily has higher performance than the other.

Lastly, as for the DX5 vs. the new DX5 II, according to their measurements and audibility thresholds considered, both devices' performance is far, far beyond a level human can discern. But technically, the ES9039Q2M chip in the DX5 II is an improved iteration over the ES9068AS (essentially feature-added version of the ES9038Q2M) in the DX5 in some measurable areas (see Ivan's 9039S development story). And as usual Topping's hardware implementation is very nice. And some useful functionalities, including PEQ, have been added to the device, too. So, IMO, if one already owns a DX5, I see no reason to upgrade. But if this is a new purchase, there is no reason to choose the DX5 over the DX5 II unless there's a huge discount (which I believe why @ICIETDIYEUR bought the DX5).

 

[simmconn]

@jkim Thanks for the background introduction. However that doesn't change where I stand. Limiting the testing bandwidth to 20KHz with a brick wall filter will not remove 2nd or 3rd harmonics of test signals between 5KHz and 6.67kHz, or 2nd harmonic of frequencies between 6.67kHz and 10kHz. So what is @staticV3's definition of 'gone completely'?

You could argue that 'Restrict BW to human hearing' does not necessarily mean a 20KHz brick wall filter, but rather a sharp cutoff filter following someone's hearing limit, or even an A-weighted result. We sure could use some clarifications there.

Due to oversampling and high order of and noise shaping, the quantization noise starts to show up in the noise floor way above 2x even 4x the sample rate. Even if the measurement BW is wide enough such that the quantization noise affects THD+N readings, its impact is equal for all test frequencies. By limiting the measurement BW, the noise component is reduced, but the harmonics of test signals <10kHz are not. The result would be a more prominent THD+N uptick in those test frequencies, if you can see it already in the high BW measurement (meaning at >5kHz, THD is already dominant in THD+N). That contradict with staticV3's claim of 'gone completely'.

Lastly, if you can't (or don't think you can) tell the audible difference between two units, just say so and no more. Please don't extend that conclusion to all humans. There may be someone out there in this planet that can reliably tell the difference when you and I cannot. I'd rather stay humble and curious.

 

[jkim]

@jkim Thanks for the background introduction. However that doesn't change where I stand. Limiting the testing bandwidth to 20KHz with a brick wall filter will not remove 2nd or 3rd harmonics of test signals between 5KHz and 6.67kHz, or 2nd harmonic of frequencies between 6.67kHz and 10kHz. So what is @staticV3's definition of 'gone completely'?

You could argue that 'Restrict BW to human hearing' does not necessarily mean a 20KHz brick wall filter, but rather a sharp cutoff filter following someone's hearing limit, or even an A-weighted result. We sure could use some clarifications there.

Now I understand precisely what your initial reply meant. I was not aware your focus was on 'technical accuracy' of that comment. But a good reader could easily see that the point of @staticV3 was not at "5kHz" or "completely gone". His main point was, the THD+N rise above a certain high frequency is (not only already low but also) beyond human hearing and hence not a concern, nor a reason why @ICIETDIYEUR should rate the device below the other. I am pretty sure in his quick write-up his focus was not on its technical accuracy.

Due to oversampling and high order of and noise shaping, the quantization noise starts to show up in the noise floor way above 2x even 4x the sample rate. Even if the measurement BW is wide enough such that the quantization noise affects THD+N readings, its impact is equal for all test frequencies. By limiting the measurement BW, the noise component is reduced, but the harmonics of test signals <10kHz are not. The result would be a more prominent THD+N uptick in those test frequencies, if you can see it already in the high BW measurement (meaning at >5kHz, THD is already dominant in THD+N). That contradict with staticV3's claim of 'gone completely'.

I definitely see your reasoning behind this. But don't you realize that your argument also in part relies on assumptions (without showing data)? First, a rise of wideband THD+N measurements between 5k - 10kHz could be substantially affected by tall order (> 3rd order) harmonics excited above 20 kHz---we do occasionally see considerable tall-order harmonics from devices. Second, in theory, fundamental tones above a certain high frequency can increase the noise level of a substantial frequency range---although I doubt this happens with 5k - 20kHz tones, some devices behave in such a way that certain frequency tones interfere with noise shaping.

Lastly, if you can't (or don't think you can) tell the audible difference between two units, just say so and no more. Please don't extend that conclusion to all humans. There may be someone out there in this planet that can reliably tell the difference when you and I cannot. I'd rather stay humble and curious.

This is a fair point. But at the same time we want to be careful in making a statement of this kind. This is not a simple issue we can discuss in one short sentence. At an extreme, as you know, we must stay humble because everyone's experience is its own and cannot be explained by theory and measurements, which challenges the main agenda of this Audio Science Review forum.

 

[ICIETDIYEUR]

Hi

My argument necessarily related to technical precision since this is an obvious observation when comparing the measurements of the two devices (DX5 and DX5 II).

It is very clear that the DX5 II reproduces frequencies beyond 5kHz with more distortion (with an additional roll-off applied to reduce this effect) than the previous model (DX5).

The bandwidth to be restored in hifi is 'normalized' between 20Hz and 20kHz, it's not me who says that: we are ok ?

So if we judge these two devices from a 'strictly scientific' point of view by comparing the distortion measurements then on the so-called 'Hi-fi' band (especially from 5kHz), the DX5 reproduces the latter better than the DX5II: there is nothing to say, develop or question because the measurements speak for themselves...

... if we say we are objective then we must remain so even when it does not suit us

Sincerely.

 

[Bleib]

It is very clear that the DX5 II reproduces frequencies beyond 5kHz with more distortion (with an additional roll-off applied to reduce this effect) than the previous model (DX5).

Someone should do the math but both DACs should be tons cleaner than the speakers. Or any music played back

 

[simmconn]

Now I understand precisely what your initial reply meant. I was not aware your focus was on 'technical accuracy' of that comment. But a good reader could easily see that the point of @staticV3 was not at "5kHz" or "completely gone". His main point was, the THD+N rise above a certain high frequency is (not only already low but also) beyond human hearing and hence not a concern, nor a reason why @ICIETDIYEUR should rate the device below the other. I am pretty sure in his quick write-up his focus was not on its technical accuracy.

I guess I don't belong to the 'good reader' then. The "5kHz" is very specific, and "completely gone" means absolute. There is really not much else in there for me to look the other way, no matter how you spin it. And I can't help but wonder, why are you working so hard to clean after him, why wouldn't he come back here and defend himself?

I definitely see your reasoning behind this. But don't you realize that your argument also in part relies on assumptions (without showing data)? First, a rise of wideband THD+N measurements between 5k - 10kHz could be substantially affected by tall order (> 3rd order) harmonics excited above 20 kHz---we do occasionally see considerable tall-order harmonics from devices. Second, in theory, fundamental tones above a certain high frequency can increase the noise level of a substantial frequency range---although I doubt this happens with 5k - 20kHz tones, some devices behave in such a way that certain frequency tones interfere with noise shaping.

I gave you a generic analysis, you respond with uncommon situations. Do you have data showing the DX5 has any of the behaviors you listed here?

I have no position in the dispute between your friend and the other member, but outright accusing someone having an 'irrational agenda' seems out of the line to me, not to mention based on some technical claim that has a weak ground.

 

[ICIETDIYEUR]

Hi.

You can read everywhere on the net: "...The fundamental frequencies are always complemented by higher harmonics that contribute to the definition of the instrument and its timbre. Much of the richness of instruments is found in the harmonics. High frequencies are therefore necessary...."

5 kHz is defined as the 'limit' between the high-mids and the highs, all frequencies from this point are very important (up to the limit of the threshold of human hearing, of course), especially if we take into account the previous sentence unless the speakers used do not have a tweeter or a very bad full-range driver...
... 'curiously' the new TOPPING DX5 II has PEQ settings beyond 5kHz: is it useless?

That said, we wonder why some people detect audible differences between devices and others don't, so do we observe the same (or the right) phenomena and interpret them correctly?

Generally, without any other explanations (or understanding) we reassure ourselves by saying that it is 'simply' psychoacoustics: it fixes 'things' well...

 

[jkim]

I gave you a generic analysis, you respond with uncommon situations.

Not at all. The harmonics don't even have to be higher-order than 3rd, if you want to relax a bit your criticism of @staticV3's cursory comment. Most modern DACs' balanced output exhibits slightly greater (or at least as strong) 3rd-order harmonics than 2nd-order ones (due to CMRR cancellation of even-order harmonics). As you know, 3rd-order harmonics are beyond human hearing when fundamentals are above 6.7 kHz. Actually, below is what he meant (and I meant):

These are the Topping D90SE's THD+N vs frequency measurements in two BW settings (solid lines = BW 20kHz, dashed lines = BW 90kHz), and for all its supported reconstruction filters (Modes 5, 6 & 7 are ones with good attenuation). With good filters (commonly chosen by most reviewers and users) and in BW 90kHz, THD+N is beginning to rise from 5k - 7kHz and elevated onwards. In BW 20kHz, however, THD+N in that range drops abruptly because the 3rd order harmonics, big contributors to BW90k THD+N, are excluded when the fundamental reaches 6.7 kHz. Sure, if you choose to, you may still want to be nitpicking about the '5 kHz' (for inaccuracy) or the 'completely gone' (for hyperbole) part in his communication. Of course, I did understand when you said that in BW 20kHz the THD would rather stand out above lower noise because the slightly rising BW90k THD+N is a sign of rising THD. But that is simply because noise is no longer the dominating factor in BW 20k THD+N. The THD is in fact still vanishingly low.

Now, as for the DX5, see the L7 Audio measurements (must've been made in the same conditions):

Essentially the same phenomenon, but in the case of DX5, there's something going on around 5kHz causing THD to rise in a narrow band.

And Amir's THD+N vs frequency measurements of the DX5 II:

Here, of course, we do not have BW 20kHz THD+N results, but can easily conjecture that the rise above 7 kHz would be gone in BW 20kHz and behave better than the DX5 because there's no peculiar peak like the DX5's. Even so, we want to be careful when comparing measurements made by different reviewers. As mentioned earlier, Wolf at L7 Audio uses the performance mode of his APx555 and full-scale sinusoids for THD+N vs frequency tests. Amir's test condition is different---for example, he uses -6 dB signals for his THD+N vs frequency measurements (the -2 dB volume in the DX5 II graph was only referring to the device volume setting).

And more critically, all this discussion is a moot point when considering any good headphones & speakers' THD levels should be at least 30 dBs higher and mask the THD of these modern electronics'. Actually, I don't see a problem objectively denying that "someone out there in this planet can reliably tell the difference," until I see the results of controlled experiments demonstrating this level of THD differences of modern DACs can be distinguishable.

I have no position in the dispute between your friend and the other member, but outright accusing someone having an 'irrational agenda' seems out of the line to me, not to mention based on some technical claim that has a weak ground.

I don't have a close relationship with @staticV3, nor any particular reason to advocate him---I never exchanged an extensive communication with him. He is just a major contributor to ASR we often see (in fact, Grand Contributor ).

 

[ICIETDIYEUR]

Hi.

Your analysis is flawed: go see what Amirm says in the post about the DX5 II and you'll understand (regarding the reason for the 90kHz BW measurements).

Don't try at all costs to convince people with what you've interpreted and believe to be correct: the 'laws' of sound reproduction are the same for all devices, and the TOPPING DX5 II, with the measurements taken by Amirm, is no exception, even if it seems to bother you

TOPPING has attempted to utilize the internal functions of the XMOS 316 chip; this isn't a DIRAC, which is something else entirely...
...if you know how to use it properly.

 

[simmconn]

Not at all. The harmonics don't even have to be higher-order than 3rd, if you want to relax a bit your criticism of @staticV3's cursory comment. Most modern DACs' balanced output exhibits slightly greater (or at least as strong) 3rd-order harmonics than 2nd-order ones (due to CMRR cancellation of even-order harmonics). As you know, 3rd-order harmonics are beyond human hearing when fundamentals are above 6.7 kHz. Actually, below is what he meant (and I meant):

What do you mean by 'Not at all'? You used two uncommon situations as counterexamples to my point ('First' and 'Second' below), but neither is reflected in the charts you re-posted. If you want to prove your point, at least come up with some data, if it is not already available somewhere on the internet.

First, a rise of wideband THD+N measurements between 5k - 10kHz could be substantially affected by tall order (> 3rd order) harmonics excited above 20 kHz---we do occasionally see considerable tall-order harmonics from devices. Second, in theory, fundamental tones above a certain high frequency can increase the noise level of a substantial frequency range---although I doubt this happens with 5k - 20kHz tones, some devices behave in such a way that certain frequency tones interfere with noise shaping.

With good filters (commonly chosen by most reviewers and users) and in BW 90kHz, THD+N is beginning to rise from 5k - 7kHz and elevated onwards. In BW 20kHz, however, THD+N in that range drops abruptly because the 3rd order harmonics, big contributors to BW90k THD+N, are excluded when the fundamental reaches 6.7 kHz. Sure, if you choose to, you may still want to be nitpicking about the '5 kHz' (for inaccuracy) or the 'completely gone' (for hyperbole) part in his communication. Of course, I did understand when you said that in BW 20kHz the THD would rather stand out above lower noise because the slightly rising BW90k THD+N is a sign of rising THD. But that is simply because noise is no longer the dominating factor in BW 20k THD+N.

Why would I relax my criticism? Even you admit that '5 kHz' is inaccurate and 'completely gone' is hyperbole. Those essentially make the claim false, simple as that.
Why are you arguing with me? The chart you re-posted and your analysis came to the same conclusion as I did. Should I thank you for agreeing with me? No, you are welcome.

And more critically, all this discussion is a moot point when considering any good headphones & speakers' THD levels should be at least 30 dBs higher and mask the THD of these modern electronics'. Actually, I don't see a problem objectively denying that "someone out there in this planet can reliably tell the difference," until I see the results of controlled experiments demonstrating this level of THD differences of modern DACs can be distinguishable.

Every time a new product came out, a seemingly thorough but same-o-same-o suite of measurement were done and then the conclusion was made that the distortion is vanishingly low and the sound this product produces should be indistinguishable from another. Yet some people tell you they sound different.

The THD is just a number, more importantly is what it reflects. To understand that, you would need more than a computer with an internet connection, listen to a lot of different products, and better yet, have an APX connected to your computer.