Topping E30 DAC Review

[Tup3x]

(I need to do some cleaning... Dust everywhere. )

Got it up and running. I hear absolutely no difference between it and my AE-5 which was excepted. Or if I would put it differently, it doesn't sound worse. This is good since it means I can't hear any noise and stuff like that i.e. sound clean (unlike my display or laptop). I can finally get proper sound when using my work laptop at home. My O2 is probably the bottleneck anyway.

I wish the Windows volume control would be "wired" to the DAC (when using USB) like it is with AE-5. It would definitely require custom driver (haven't installed one and not going to if the only benefit is ASIO driver). Currently using the DAC mode and controlling the volume through Windows 10 since I don't have batteries for the remote control yet -_-. I'll have to be careful to not use WASAPI exclusive mode...

And when it comes to "Velvet Sound" vs. "Sabre glare"... It's all utter bullshit assuming the implementations are even half decent (it certainly is possible to ruin every DAC with incompetent engineering) and used filter even somewhat similar.

Then I have to send my mom to Finland She's Finnish and can travel in these Corona times.

Hifitalo in Salo has them.

 

[voodooless]

Just made a first setup for my home office using old modified Hypex UcD400 amp and LS50. I let OSX upsample to 32bit/384kHz. Seems to work just fine even upto 32/768. Can't really turn up the volume at the moment, but it sounds like a decent little DAC/pre

 

[samsa]

I let OSX upsample to 32bit/384kHz. Seems to work just fine even upto 32/768.

Just curious. What does upsampling do for you?

 

[voodooless]

Just curious. What does upsampling do for you?

It gives nicer square waves .. otherwise, probably not so much. The DAC will probably have more stop band attenuation due to the software filtered being more capable. Possibly Less pre and post ringing. Is it audible? probably not. However when there are multiple sampling rates busy on a system, one best picks something fairly high. It’s more out of convenience

 

[samsa]

The DAC will probably have more stop band attenuation due to the software filtered being more capable. Possibly Less pre and post ringing.

The characteristics of the reconstruction filters on the AK4493 are published (Archimago measured the impulse response). So the "probably" and "possibly" can be checked.

I am really quite curious whether the performance of software upsampling actually exceeds that of the hardware filters on modern AKM chips.

Is it audible? probably not.

I don't care whether it's audible; I just want to know whether it's better.

(The cognoscenti complain that AKM's "sharp rolloff" filter is anything but. However, since you talk about less pre- and post-ringing, I assume you are not using a sharper filter than AKM offers.)

 

[JohnYang1997]

It gives nicer square waves .. otherwise, probably not so much. The DAC will probably have more stop band attenuation due to the software filtered being more capable. Possibly Less pre and post ringing. Is it audible? probably not. However when there are multiple sampling rates busy on a system, one best picks something fairly high. It’s more out of convenience

Upsamplig doesn't automatically give you
nicer square wave. In contrast, ideal upsampling gives exactly same result as original sampling rate in digital domain. You get to choose the filter to do the upsampling.
Better attenuation, sharper filter = more "ringing" (which is ideal but you may lose digital headroom or clip in the dac)
Pretty square wave = poor attenuation (very non ideal as you gets imaging and distortion)

 

[JohnYang1997]

The characteristics of the reconstruction filters on the AK4493 are published (Archimago measured the impulse response). So the "probably" and "possibly" can be checked.

I am really quite curious whether the performance of software upsampling actually exceeds that of the hardware filters on modern AKM chips.



I don't care whether it's audible; I just want to know whether it's better.

(The cognoscenti complain that AKM's "sharp rolloff" filter is anything but. However, since you talk about less pre- and post-ringing, I assume you are not using a sharper filter than AKM offers.)

AKM's filters are leaky. But shouldn't be much of a difference. Sharp roll off should be close to ideal anyway.

 

[MadMan]

Is there any reason you're skipping frequency response for a little while now @amirm ? I understand it might seem to be a 'trivial' solved problem, but the trend toward weaker reconstruction filters, perhaps for better measurements, used to lead to some (small) high frequency rolloff. I'd at least like to know if companies are making this tradeoff, even if it's limited by what's available from the DAC chip.

 

[JohnYang1997]

Is there any reason you're skipping frequency response for a little while now @amirm ? I understand it might seem to be a 'trivial' solved problem, but the trend toward weaker reconstruction filters, perhaps for better measurements, used to lead to some (small) high frequency rolloff. I'd at least like to know if companies are making this tradeoff, even if it's limited by what's available from the DAC chip.

It's Not skipped and clearly shown in the review.

 

[MadMan]

It's Not skipped and clearly shown in the review.

No, it's not. There is: dashboard summary, SINAD comparison chart, dynamic range, IMD, multitone, jitter, THD v frequency, FFT for filters (18kHz and up), and linearity. Am I missing something from one of those ? Inferring the frequency response from one of those graphs?

 

[JohnYang1997]

No, it's not. There is: dashboard summary, SINAD comparison chart, dynamic range, IMD, multitone, jitter, THD v frequency, FFT for filters (18kHz and up), and linearity. Am I missing something from one of those ? Inferring the frequency response from one of those graphs?

This is frequency response for high frequencies.

 

[samsa]

AKM's filters are leaky. But shouldn't be much of a difference. Sharp roll off should be close to ideal anyway.

AKM's "sharp rolloff" is as slow as you can get (transition band extends to 24.1kHz) without aliasing creeping into the audio band. A sharper ("brickwall") rolloff is one thing that — if that was what you were after — you would need to implement in software. But @voodooless said he wanted less ringing (and "better square waves"), so it's pretty clear that sharper rolloff is not what he is after.

The stop-band attenuation of AKM's not-so-sharp "sharp rolloff" is 100dB, which seems excellent to me.

So I'm curious what combination of filter characteristics @voodooless is trying to achieve with software upsampling.

 

[mshenay]

So for what it's worth subjectively the big thing I notice different from the D10 to E30 [using Filter 1] is "depth" or echo, listening to a nice copy of "Hell on Heels" from the Pistol Annies and there's a distinct but slight and very quiet echo of their voices. It's not really discernible with D10 [or Modi 3] but it's quite apparent with E30

Now my Subjective comparisons to D10 are down with a PreAmp that's hot fed with both E30 and D10 line outs, so flip a switch and listen. An yea I do em double blind with the assistance of my spouse

Double blind aside, D10 has always sounded flat to me as most often recordings where I'm used to hearing that slight echo in the room or sometimes reflections off walls [on less than perfect masters but "good" recordings"] are most often missing...

Non the less given the price difference I certainly think E30 is quite the steal! Assuming you DON'T Need USB to Coax conversion

 

[JohnYang1997]

AKM's "sharp rolloff" is a slow as you can get (transition band extends to 24.2kHz) without aliasing creeping into the audio band. A sharper ("brickwall") rolloff is one thing that — if that was what you were after — you would need to implement in software. But @voodooless said he wanted less ringing (and "better square waves"), so it's pretty clear that sharper rolloff is not what he is after.

The stop-band attenuation of AKM's not-so-sharp "sharp rolloff" is 100dB, which seems excellent to me.

So I'm curious what combination of filter characteristics @voodooless is trying to achieve with software upsampling.

That's right. As I said AKM's filters are leaky. Also mentioned in the previous post above the one you read. Sharp filter is ideal. The ringing is expected behavior and good. The only problem it gives is digital clipping. Better looking square wave doesn't improve anything in the audio band but due to leaky filter the imaging are much worse appearing in the audio band. It is probably worse. Not potentially better.

 

[voodooless]

The characteristics of the reconstruction filters on the AK4493 are published (Archimago measured the impulse response). So the "probably" and "possibly" can be checked.

Im my case it cannot as far as I can see. While we do have the specs of the AKM filters, I cannot find any recent info on OSX'es. You can find some older info, which might be accurate still today, but it might also not be. That will give about 100dB of attenuation. About the same as the AK filters.

I am really quite curious whether the performance of software upsampling actually exceeds that of the hardware filters on modern AKM chips.

If you look at some dedicated software resamping lib like Sox, you can conclude that they can do a better job than the DAC filters. It will give you up to 170 dB of stopband attenuation from 22 Khz already. That is definitely better performance than most (if not all) DAC chips ofer. Here you can compare various software options:

(The cognoscenti complain that AKM's "sharp rolloff" filter is anything but. However, since you talk about less pre- and post-ringing, I assume you are not using a sharper filter than AKM offers.)

No, I'm using filter 1 comfortably

Upsamplig doesn't automatically give you
nicer square wave. In contrast, ideal upsampling gives exactly same result as original sampling rate in digital domain. You get to choose the filter to do the upsampling.
Better attenuation, sharper filter = more "ringing" (which is ideal but you may lose digital headroom or clip in the dac)
Pretty square wave = poor attenuation (very non ideal as you gets imaging and distortion)

Hmm, yes, you are right. Removing more high frequency components would make the square wave actually look "worse". I've seen a post somewhere a few days ago that seemed to indicate that that is not always the case.. can't find it anymore though.. And logically thinking about it.. it cannot be.

So for what it's worth subjectively the big thing I notice different from the D10 to E30 [using Filter 1] is "depth" or echo, listening to a nice copy of "Hell on Heels" from the Pistol Annies and there's a distinct but slight and very quiet echo of their voices. It's not really discernible with D10 [or Modi 3] but it's quite apparent with E30

Is the echo gone in any of the other filter modes on the E30?

 

[mshenay]

Is the echo gone in any of the other filter modes on the E30?

Never quite gone but the non sharp filters it's not as discernible, where as with 1 it's rather apparent and noticeable when hot swapping between D10 and E30

I tend not to listen to AKM D/S without a sharp filter so I didn't really explore much

 

[Axaion]

I havent noticed any echo yet on mine, playing games, voice chat.. music (not super high quality), videos.. also on F-1 Sharp

 

[voodooless]

Never quite gone but the non sharp filters it's not as discernible, where as with 1 it's rather apparent and noticeable when hot swapping between D10 and E30

If it’s an actual echo you hear, it’s probably not the reconstruction filter. Pre echo’s are reversed in time and usually only audible in high transients, if ever. The ringing of the reconstruction filters is usually very high in frequency anyway. And post-echoes are usually masked by the main signal.

 

[mshenay]

If it’s an actual echo you hear, it’s probably not the reconstruction filter. Pre echo’s are reversed in time and usually only audible in high transients, if ever. The ringing of the reconstruction filters is usually very high in frequency anyway. And post-echoes are usually masked by the main signal.

Hmm let me clarify, the echo I'm hearing is recorded the sound of the voices I'm not talking about an audible digital artifact,

 

[samsa]

Hmm let me clarify, the echo I'm hearing is recorded the sound of the voices I'm not talking about an audible digital artifact,

So you're saying that the E30 is allowing you to hear things in the recording that the D10 did not?