Topping E30 II Lite DAC review and measurements

[Billy Budapest]

I mean already for $90 it shows that DAC's don't need to be expensive to be good.

Right. I am just saying that when the quality filters down to the $9 level, then nobody will be able to debate that DACs are a solved problem at all price levels.

 

[Jimster480]

Right. I am just saying that when the quality filters down to the $9 level, then nobody will be able to debate that DACs are a solved problem at all price levels.

True that, I think the apple or Google dongle was already pretty close though AFAIK?

 

[natna]

Wow. A lot of work man! Nice job!

For most people, who are streaming from a music platform or playing hi-res files on PC/phone/tablet, it's a great unit. Almost a technical miracle.
Unfortunately for guys like me, that want to use it with other sources, like older CD/BD players, the SPDIF input performance is a no go.

 

[Rja4000]

Wow. A lot of work man! Nice

Thanks.

for guys like me, that want to use it with other sources, like older CD/BD players, the SPDIF input performance is a no go.

The Jitter on SPDIF is not how we like it, for sure, but there is very little chance it would make any audible difference.

Some critical measurements above, like the Multitone measurement, where perfect sync between the DAC and RME clock is required, were performed using the SPDIF input.

As you can see, there is nothing to complain about.

And, anyway, the result will still outperform the capabilities of the older sources you want to connect.

So that should not stop you to use this one, IMO.
As it doesn't stop me.

 

[morillon]

Thanks.

The Jitter on SPDIF is not how we like it, for sure, but there is very little chance it would make any audible difference.

Some critical measurements above, like the Multitone measurement, where perfect sync between the DAC and RME clock is required, were performed using the SPDIF input.

As you can see, there is nothing to complain about.

And, anyway, the result will still outperform the capabilities of the older sources you want to connect.

So that should not stop you to use this one, IMO.
As it doesn't stop me.

could you explain a little the effort of "synchro" of dac rme that you had to make for this multitone etc?
thank you..

 

[Rja4000]

could you explain a little the effort of "synchro" of dac rme that you had to make for this multitone etc?
thank you..

Sure.

Multitone is a test Audioprecision is offering.
You may find the description here.

For this test to give full benefits, they generate a signal -here with 32 tones separated by approximately 1/3 octave- where each "peak" falls exactly in one bin of the FFT.

AudioPrecision has a way to align timing within the Analyzer, to resync the signal with the FFT engine, but the software I'm using (and no other software I'm aware of) doesn't have that.
Therefore, the only way to make this happen is (in this case) to send the clock to the DAC from the ADC.
Then, a "no window" FFT transform may be used with clear separation of signal and Noise+Distortion.

If the sync doesn't work, "no window" FFT becomes unusable. One then rely on "Blackman-Harris 7" or other highly discriminant FFT window, but they have wider lobes that won't resolve the very close low frequencies.

(FFT bins are at a constant frequency distance, while tones are spaced following a logarithmic law. So the number of bins between successive tones is very low at low frequencies and very big at highest ones.)

Here is a comparison when it works

and when it doesn't
(From the Fiio BTA30 Pro review)

I use Audioprevision APx500 Flex software in Demo mode to generate the exact AP 32 tones at different frequencies and FFT sizes.

It's important to note that TD+N varies with the crest factor of the file, which varies between 44.1kHz multiples and 48kHz multiples.
(Probably a reason amongst others Amir doesn't use this figure.)



There are other cases where this is very helpful:
One may get the frequency response in a second by using REW Periodic noise, as an example. But you also have to use "no window" and to be in sync for that to work.

Also, for linearity, I prefer to use the "no window" FFT since it's more accurate for level measurements and allow easily to control the width you're considering for your signal.

 

[morillon]

Bien sûr.

Multitone est un test proposé par Audioprecision.
Vous pouvez trouver la description ici .

Pour que ce test donne tous les bénéfices, ils génèrent un signal -ici avec 32 tonalités séparées d'environ 1/3 d'octave- où chaque "pic" tombe exactement dans une case de la FFT.

AudioPrecision dispose d'un moyen d'aligner la synchronisation dans l'analyseur, de resynchroniser le signal avec le moteur FFT, mais le logiciel que j'utilise (et aucun autre logiciel à ma connaissance) ne l'a pas.
Par conséquent, la seule façon d’y parvenir est (dans ce cas) d’envoyer l’horloge au DAC depuis l’ADC.
Ensuite, une transformation FFT « sans fenêtre » peut être utilisée avec une séparation claire du signal et du bruit + distorsion.

Si la synchronisation ne fonctionne pas, la FFT « sans fenêtre » devient inutilisable. On s'appuie alors sur "Blackman-Harris 7" ou une autre fenêtre FFT hautement discriminante, mais ils ont des lobes plus larges qui ne résoudront pas les basses fréquences très proches.

(Les groupes FFT sont à une distance de fréquence constante, tandis que les tonalités sont espacées selon une loi logarithmique. Ainsi, le nombre de groupes entre les tonalités successives est très faible aux basses fréquences et très grand aux plus hautes.)

Voici une comparaison quand ça marche

View attachment 309515

et quand ce n'est pas le cas
(D'après la revue Fiio BTA30 Pro )

View attachment 309516

J'utilise le logiciel Audioprevision APx500 Flex en mode Démo pour générer les tonalités AP 32 exactes à différentes fréquences et tailles FFT.

Il est important de noter que TD+N varie en fonction du facteur de crête du fichier, qui varie entre des multiples de 44,1 kHz et des multiples de 48 kHz.
(Probablement une des raisons pour lesquelles Amir n'utilise pas ce chiffre.)



Il existe d'autres cas où cela s'avère très utile :
On peut obtenir la réponse en fréquence en une seconde en utilisant le bruit périodique REW, comme exemple. Mais il faut aussi utiliser "no window" et être synchronisé pour que cela fonctionne.

De plus, pour la linéarité, je préfère utiliser la FFT "sans fenêtre" car elle est plus précise pour les mesures de niveau et permet de contrôler facilement la largeur que vous envisagez pour votre signal.

from memory, if amirm often uses a dolph 200 with the ap, I don't know if he detailed what he uses in the specific case of multitones (44k)?
Multitones from Pkanes has been offering an AP32 test for a long time. It will be necessary to look at the FFT conditions associated with it out of curiosity....
(to be honest...at the simplest...I leave aside the low frequency observation of multitones and use kaiser 12 and above, to observe "higher" ;-) ,
can be a precaution to be considered during exchanges as it is linked to the effort of protocols etc. do not show above 100-200hz... we often come across reactions on low frequencies which do not stand out... ? )

 

[Rja4000]

from memory, if amirm often uses a dolph 200 with the ap

AP is using a proprietary Window function, derived from Dolph-Chebyshev.
(Read here)

I don't know if he detailed what he uses in the specific case of multitones (44k)?

Amir once wrote he is using 192kHz sampling, 256k FFT for Multitone.
Given how narrow the peaks look, it's using Rectangle ("no window") FFT function.

Multitones from Pkanes has been offering an AP32 test for a long time.

The excellent Multitone software generates a different 32 tones, where lower frequencies are more spread appart (to allow better discrimination of low frequency peaks with usual functions I suppose).

 

[morillon]

AP is using a proprietary Window function, derived from Dolph-Chebyshev.
(Read here)

Amir once wrote he is using 192kHz sampling, 256k FFT for Multitone.
Given how narrow the peaks look, it's using Rectangle ("no window") FFT function.

The excellent Multitone software generates a different 32 tones, where lower frequencies are more spread appart (to allow better discrimination of low frequency peaks with usual functions I suppose).

in case, the ap32 in 192k has been made available here...
if you have the opportunity to make some versions in lower resolutions to avoid conversions I would be very grateful.. ;-)

 

[Rja4000]

Topping gives 20 ohm as output impedance in their specs.

I've performed a quick check

Output impedance

Open circuit Voltage	V1	2.1138​	Vrms
Load resistance	RL	613.9​	Ohm
Loaded circuit Voltage	V2	2.0454​	Vrms
Zo=RL*(V1/V2-1)	Zo	20.5​	Ohm

(Measured at 997Hz only, with a calibrated Brymen BM869s)

 

[Veri]

I've performed a quick check

Output impedance

Open circuit Voltage	V1	2.1138​	Vrms
Load resistance	RL	613.9​	Ohm
Loaded circuit Voltage	V2	2.0454​	Vrms
Zo=RL*(V1/V2-1)	Zo	20.5​	Ohm

(Measured at 997Hz only, with a calibrated Brymen BM869s)

So, comfortably low. Thanks

 

[Billy Budapest]

True that, I think the apple or Google dongle was already pretty close though AFAIK?

Not SOTA performance yet. But they are pretty decent (especially the Apple dongle) and actually the performance for the size and price is really good.

 

[Moto]

@Rja4000 what does your new 600 ohm load look like?

 

[pkane]

The excellent Multitone software generates a different 32 tones, where lower frequencies are more spread appart (to allow better discrimination of low frequency peaks with usual functions I suppose).

There are actually multiple ways to generate multitones with Multitone. A preconfigured test signal named Multitone 32 (AP) is based on both, frequencies and phase of the AP multitone signal that Amir uses.

 

[morillon]

There are actually multiple ways to generate multitones with Multitone. A preconfigured test signal named Multitone 32 (AP) is based on both, frequencies and phase of the AP multitone signal that Amir uses.

View attachment 309597

"what else?:
;-)

 

[Rja4000]

There are actually multiple ways to generate multitones with Multitone. A preconfigured test signal named Multitone 32 (AP) is based on both, frequencies and phase of the AP multitone signal that Amir uses.

View attachment 309597

... for which you then use "no window" FFT.

 

[Rja4000]

@Rja4000 what does your new 600 ohm load look like?

It looks like an XLR female connector with a metal film 620 ohm (613.9 Ohm actually) soldered between pins 2 and 3.
Should be OK up to 24dBu.

 

[pkane]

... for which you then use "no window" FFT.

Sorry, but I've not had a chance to read the rest of the messages. What is the significance of "no window" or "window" FFT in this context?

 

[antcollinet]

Right. I am just saying that when the quality filters down to the $9 level, then nobody will be able to debate that DACs are a solved problem at all price levels.

It basically is. The apple dongle achieves a sinad of just under 100dB with other measures to match.

That is audibly indistinguishable from this.

Dacs are a solved problem at all price levels.

 

[Billy Budapest]

It basically is. The apple dongle achieves a sinad of just under 100dB with other measures to match.

That is audibly indistinguishable from this.

Dacs are a solved problem at all price levels.

So now the question is, just what is the most expensive DAC on the market with comparable (or worse) performance to the Apple dongle?