Gustard A26 DAC & Streamer Review

[BDWoody]

It is hilarious that some Toping and SMLS DACs are rated above this DAC. This DAC smokes them. It really is no contest in the real world. Topping and SMLS get their high specs by using endless amounts of negative feedback. High amounts of negative feedback can have a notorious effect on sound. Also switching in a Class AB circuit does not sound as good as a well-designed Class A circuit. Hence, Gustard specs are amazing considering they designed it to also sound really good and avoid the known pitfalls of some designs.

 

[Tank81]

May i ask why is this thread full of peoples who wont even own this Dac saying that its worse than smth? I got a26, Ifi neo and Schiit bifrost 2, used to own topping d90 aswell and I can happily say that a26 sounds better than all of those. Its more fuller on bass side and got way more clarity on mid section. I used to have girl with perfect body measurements but she sounded like nagging witch. Same goes to dac and amps.

 

[voodooless]

May i ask why is this thread full of peoples who wont even own this Dac saying that its worse than smth?

Well, as a consolation, I think you’re both wrong

I got a26, Ifi neo and Schiit bifrost 2, used to own topping d90 aswell and I can happily say that a26 sounds better than all of those. Its more fuller on bass side and got way more clarity on mid section.

Did you even compare them side by side? Did you do the test with at least the most basic of controls?

I used to have girl with perfect body measurements but she sounded like nagging witch. Same goes to dac and amps.

But body measurements are not very well correlated to the words that come out of said body, contrary to what we know about electronics measurements with decades worth of research to back it up. So this comment besides being offensive, doesn’t make any sense.

 

[BDWoody]

May i ask why is this thread full of peoples who wont even own this Dac saying that its worse than smth?

I haven't seen that.

I got a26, Ifi neo and Schiit bifrost 2, used to own topping d90 aswell and I can happily say that a26 sounds better than all of those. Its more fuller on bass side and got way more clarity on mid section.

I used to have girl with perfect body measurements but she sounded like nagging witch. Same goes to dac and amps.

 

[ZestClub]

I don't stream music from a service but play it from memory card - can I use this Gustard A26 with a USB hard drive plugged-in into the USB port?
Or would I also need some other form of player.

 

[AdamFrandsen]

Help me understand this, the thd+n vs freq graph is almost flat at 0.001%/100db/16.6bits also at 1khz, yet the sinad/thd+n figure is 0.000126%/117.975db/19.6bits
Gustards own thd+n vs freq graph shows an average of -120db, so very far from -100db

 

[mdsimon2]

Help me understand this, the thd+n vs freq graph is almost flat at 0.001%/100db/16.6bits also at 1khz, yet the sinad/thd+n figure is 0.000126%/117.975db/19.6bits
Gustards own thd+n vs freq graph shows an average of -120db, so very far from -100db

Different bandwidths, dashboard is 22 kHz bandwidth and THD+N vs frequency is 90 kHz bandwidth.

Michael

 

[AdamFrandsen]

Different bandwidths, dashboard is 22 kHz bandwidth and THD+N vs frequency is 90 kHz bandwidth.

Michael

I am sorry, but I still don’t understand this. Feeding a 48khz sample should only generate up to 24khz contents according to nyquist so why 90khz bandwidth, and shouldn’t the graphs/numbers correlate up until 22khz regardless, the graph only shows until 20khz… I am confused

I also would like to see a 124 multi-tone test instead of 32 to reflect the actual number of notes between 15hz (lowest note on a Bösendorfer grand 16.35hz) and 20khz, where can one suggest this

 

[Billy Budapest]

May i ask why is this thread full of peoples who wont even own this Dac saying that its worse than smth? I got a26, Ifi neo and Schiit bifrost 2, used to own topping d90 aswell and I can happily say that a26 sounds better than all of those. Its more fuller on bass side and got way more clarity on mid section. I used to have girl with perfect body measurements but she sounded like nagging witch. Same goes to dac and amps.

It’s the power of imagination. You think you hear a difference between those DACs, but that would be impossible due to the physics of sound and human biology. All of the differences among the units in distortion and noise are under the limits of human hearing.

 

[mdsimon2]

I am sorry, but I still don’t understand this. Feeding a 48khz sample should only generate up to 24khz contents according to nyquist so why 90khz bandwidth, and shouldn’t the graphs/numbers correlate up until 22khz regardless, the graph only shows until 20khz… I am confused

I also would like to see a 124 multi-tone test instead of 32 to reflect the actual number of notes between 15hz (lowest note on a Bösendorfer grand 16.35hz) and 20khz, where can one suggest this

There are few reasons for the larger measurement bandwidth to give different THD+N results. Remember that we are looking THD + N, so noise can have a big impact even if there are no out of band harmonics. Many DACs implement noise shaping which results in rising out of band noise, for example here is a measurement from a Focusrite 18i20 2nd gen showing rising noise past 20 kHz.

Even if the noise level is flat from 22 kHz to 90 kHz, because the measurement is being taken over a larger bandwidth the cumulative noise will be greater with the larger bandwidth.

Here are a few examples from a MOTU Ultralite MK5 running at 48 kHz in to a Cosmos ADC running at 384 kHz.

First let's look at a low level signal (-60 dBFS). Here you can see that the noise level is flat from 20 Hz to 100 KHz. The spectrum doesn't change but because noise is being integrated over a larger bandwidth SNR goes down.

1) 22 kHz BW - SNR at -60 dBFS = 61.5 dB, FS SNR = 61.5 + 60 = 121.5 dB

2) 48 kHz BW - SNR at -60 dBFS = 58.6 dB, FS SNR = 58.6 + 60 = 118.6 dB

3) 90 kHz BW - SNR at -60 dBFS = 56.2 dB, FS SNR = 56.2 + 60 = 116.2 dB

Here are a few examples with the same setup but using a higher level signal (4 V). Here you can see there is now some high frequency junk showing up but it is very low in level. Again the biggest difference here is integrating noise over a larger bandwidth, SNR decreases but THD stays about the same, as a result THD+N gets worse.

1) 22 kHz BW - THD = -116.0 dB, SNR = 108.6 dB, THD+N = -107.9 dB

2) 48 kHz BW - THD = -115.4 dB, SNR = 105.3 dB, THD+N = -104.9 dB

3) 90 kHz BW - THD = -115.6 dB, SNR = 102.0 dB, THD+N = -101.8 dB

Hope this helps.

Michael

 

[AdamFrandsen]

There are few reasons for the larger measurement bandwidth to give different THD+N results. Remember that we are looking THD + N, so noise can have a big impact even if there are no out of band harmonics. Many DACs implement noise shaping which results in rising out of band noise, for example here is a measurement from a Focusrite 18i20 2nd gen showing rising noise past 20 kHz.

Even if the noise level is flat from 22 kHz to 90 kHz, because the measurement is being taken over a larger bandwidth the cumulative noise will be greater with the larger bandwidth.

Here are a few examples from a MOTU Ultralite MK5 running at 48 kHz in to a Cosmos ADC running at 384 kHz.

First let's look at a low level signal (-60 dBFS). Here you can see that the noise level is flat from 20 Hz to 100 KHz. The spectrum doesn't change but because noise is being integrated over a larger bandwidth SNR goes down.

1) 22 kHz BW - SNR at -60 dBFS = 61.5 dB, FS SNR = 61.5 + 60 = 121.5 dB
View attachment 354820

2) 48 kHz BW - SNR at -60 dBFS = 58.6 dB, FS SNR = 58.6 + 60 = 118.6 dB
View attachment 354822

3) 90 kHz BW - SNR at -60 dBFS = 56.2 dB, FS SNR = 56.2 + 60 = 116.2 dB
View attachment 354823

Here are a few examples with the same setup but using a higher level signal (4 V). Here you can see there is now some high frequency junk showing up but it is very low in level. Again the biggest difference here is integrating noise over a larger bandwidth, SNR decreases but THD stays about the same, as a result THD+N gets worse.

1) 22 kHz BW - THD = -116.0 dB, SNR = 108.6 dB, THD+N = -107.9 dB
View attachment 354817

2) 48 kHz BW - THD = -115.4 dB, SNR = 105.3 dB, THD+N = -104.9 dB
View attachment 354818

3) 90 kHz BW - THD = -115.6 dB, SNR = 102.0 dB, THD+N = -101.8 dB
View attachment 354819

Hope this helps.

Michael

It makes sense the way you show me that the increase in noise at higher frequencies is factored in, when you test over a larger bandwidth. It also shows up to 100khz. The graph I cannot understand is the one below, which shows a flat curve across all frequencies at thd+n of 0.001% for a dac that has a SINAD of 117db, that is nearly the same as the Brigasti, which was just tested at a SINAD of 83db (if I remember right), no other DAC on here has this large discrepancy. I can see there is some jitter noise, the reason I am looking at this specific unit, is because it is more or less the same as the Audalytic ah90 with better power supply. However the ah90 has a CK01 jitter clock module which may relieve it of this problem? It also has option for external linear power supply… Anyhow, wish I could understand how the a26 suddenly degrades so much…

 

[mdsimon2]

It makes sense the way you show me that the increase in noise at higher frequencies is factored in, when you test over a larger bandwidth. It also shows up to 100khz. The graph I cannot understand is the one below, which shows a flat curve across all frequencies at thd+n of 0.001% for a dac that has a SINAD of 117db, that is nearly the same as the Brigasti, which was just tested at a SINAD of 83db (if I remember right), no other DAC on here has this large discrepancy. I can see there is some jitter noise, the reason I am looking at this specific unit, is because it is more or less the same as the Audalytic ah90 with better power supply. However the ah90 has a CK01 jitter clock module which may relieve it of this problem? It also has option for external linear power supply… Anyhow, wish I could understand how the a26 suddenly degrades so much…

There are other DACs which show similar discrepancies between the 1 kHz THD+N dashboard measurement and the wideband THD+N sweep. In cases where there is no frequency dependency in the THD+N sweep, that indicates there is some elevated out of band noise.

Sometimes Amir shows a wideband FFT to help confirm. See this review as an example -> https://www.audiosciencereview.com/forum/index.php?threads/smsl-do200-mkii-dac-review.41396/, it goes from -120 dB THD+N in the dashboard to -97 dB in the wideband THD+N sweep and the wideband FFT clearly shows noise shaping.

Michael

 

[El_Arte]

There are few reasons for the larger measurement bandwidth to give different THD+N results. Remember that we are looking THD + N, so noise can have a big impact even if there are no out of band harmonics. Many DACs implement noise shaping which results in rising out of band noise, for example here is a measurement from a Focusrite 18i20 2nd gen showing rising noise past 20 kHz.

Even if the noise level is flat from 22 kHz to 90 kHz, because the measurement is being taken over a larger bandwidth the cumulative noise will be greater with the larger bandwidth.

Here are a few examples from a MOTU Ultralite MK5 running at 48 kHz in to a Cosmos ADC running at 384 kHz.

First let's look at a low level signal (-60 dBFS). Here you can see that the noise level is flat from 20 Hz to 100 KHz. The spectrum doesn't change but because noise is being integrated over a larger bandwidth SNR goes down.

1) 22 kHz BW - SNR at -60 dBFS = 61.5 dB, FS SNR = 61.5 + 60 = 121.5 dB
View attachment 354820

2) 48 kHz BW - SNR at -60 dBFS = 58.6 dB, FS SNR = 58.6 + 60 = 118.6 dB
View attachment 354822

3) 90 kHz BW - SNR at -60 dBFS = 56.2 dB, FS SNR = 56.2 + 60 = 116.2 dB
View attachment 354823

Here are a few examples with the same setup but using a higher level signal (4 V). Here you can see there is now some high frequency junk showing up but it is very low in level. Again the biggest difference here is integrating noise over a larger bandwidth, SNR decreases but THD stays about the same, as a result THD+N gets worse.

1) 22 kHz BW - THD = -116.0 dB, SNR = 108.6 dB, THD+N = -107.9 dB
View attachment 354817

2) 48 kHz BW - THD = -115.4 dB, SNR = 105.3 dB, THD+N = -104.9 dB
View attachment 354818

3) 90 kHz BW - THD = -115.6 dB, SNR = 102.0 dB, THD+N = -101.8 dB
View attachment 354819

Hope this helps.

Michael

Excellent explanation! Congratulations and thank you!

 

[AdamFrandsen]

There are other DACs which show similar discrepancies between the 1 kHz THD+N dashboard measurement and the wideband THD+N sweep. In cases where there is no frequency dependency in the THD+N sweep, that indicates there is some elevated out of band noise.

Sometimes Amir shows a wideband FFT to help confirm. See this review as an example -> https://www.audiosciencereview.com/forum/index.php?threads/smsl-do200-mkii-dac-review.41396/, it goes from -120 dB THD+N in the dashboard to -97 dB in the wideband THD+N sweep and the wideband FFT clearly shows noise shaping.

Michael

Thanks for taking the time Michael! What is often the reason for such out of band noise?

 

[mdsimon2]

Thanks for taking the time Michael! What is often the reason for such out of band noise?

I am far from an expert in this area but noise shaping is a common technique to improve audio band SNR by shifting noise out of band. A bit of googling or searching ASR for DAC noise shaping will give you more info, some rather simple and some very complicated. As a layman I certainly don't understand all of the technical details behind it.

Michael

 

[anton123]

How is it working with Roon? Any issues? I can't find Gustard in Partners list and any confirmations anywhere ...

 

[Snoopy]

How is it working with Roon? Any issues? I can't find Gustard in Partners list and any confirmations anywhere ...

I believe it only works like a raspberry pi would for roon.

 

[anton123]

I've bought A26, but got a problem. When I'm playing files 16bit 44.1khz via Roon - I can hear noise, the best way to hear it - put amp volume to 0 and I hear this noise, but in the same time when I'm playing file 96khz or 192khz - there is no any noise. And when Ipaused playback - there is no noise... Maybe someone already met with this problem? Any suggestions?

 

[piercer]

Long time lurker here. Based on the review/analysis and general opinion of people here I went ahead and imported an A26 from China. I use Roon and could not be happier going from a 2-box to one-box solution. I had a chord Qutest and Stack Audio Link before and I honestly think the A26 sounds better. Very happy with it.

I would like to congratulate the HiFiCat store for doing an exellent job shipping it from China - really impressed with the transparency of the tracking, unlike some things I have ordered from other companies.