@Taz777
Thank you for continuing to share your experiences with the A18. You're sure putting that DAC through its paces. That's excellent!
About the DSD limitation you found with Audirvana, that makes sense. I saw PCM and DSD limitations too when testing Volumio on RPi, but none of them relevant to real-world usage. Anything above PCM 192 KHz or DSD256 really is just for testing, not practical delivery formats for music. Even among those those want to upsample in software I've seen comments that DSD512 actually hurt more than it helped -- in addition to using obscene amounts of CPU (if upsampling on the fly) or disk space (if stored).
I was glad to see that you found I2S on the A18 to be pin compatible with your Cayin N6ii, as I understand it isn't standardized and can otherwise require custom cables.
But I am VERY surprised that you are hearing differences between USB and I2S on the same DAC, since both are feeding a digital stream to the same DAC chip and playing through the same output stage. The explanations I can think of are:
1) Difference in jitter from the different input circuitry and/or phase shift caused by I2S using an embedded clock signal from the source. Amir on ASR measured I2S on a Gustard X26 two years ago, to dispel the myth that I2S was inherently superior. He found significant phase shift in one of the channels on I2S. It could be just the units he tested or whatever, but it's definitely not a good thing, even if it isn't audible. (Let's not get into the war between extremists of ASR and head-fi. That's not my point here. I believe in both.)
2) Difference in the digital source: If one (say USB) is bit-perfect (being upsampled by the DAC chip) and the other (say I2S) is being upsampled by the source (in this case by the Cayin) then that could definitely cause a difference, as they are likely using different upsampling algorithms. I would be interested if you can still hear a difference if both inputs are fed 16/44.
3) Different MQA processing: I am surprised if the Cayin can do a full unfold of MQA and pass it digitally to the Gustard. I would have thought it could only perform a first unfold and pass 88.2/96KHz when working as a decoder. If so, the I2S feed would only have a partial unfold by the Cayin and no MQA filter applied. (The A18 can only process MQA through USB, as you know). Some people will argue that the less MQA the better, but that's besides the points for this discussion, as there could (or rather: should) be a difference between USB and I2S caused by this when playing MQA. Even if the Cayin can perform a full unfold (say to 352.8KHz as shown on your display) there still would be no MQA filter used by the Gustard when fed through I2S.
As for the difference between Topping D90 and Gustard A18, I much appreciate your posting about this.
I am skeptical that D90's XU208 vs A18's XU216 would make an audible difference, since (again) it's a digital signal, which presumably arrives intact into the same AK4499 in both DACs. But I guess slight differences in jitter are possible.
The output stages are far more likely to cause a difference. AK4499 outputs an insane amount of current (arguably too much) which has to be tamed and converted into voltage (I/V) in the output stage. Not a trivial task. D90 gets it down to 4Vrms (and provides an XLR/RCA switch) whereas A18 lets through a full 6.3Vrms (and no switch). And still, D90 has better THD+N, which makes me think Topping simply spent a bit more on their output stage.
But that's all speculation. I am no electronics expert anyway
Edit: OTOH, Topping was the first vendor to focus squarely on measurements, essentially using ASR as primary channel to promote their gear. And they have been rewarded with valuable recommendations from Amir of their affordable, yet well measuring products. Conversely, I remember seeing quotes from Gustard (sorry, I forget where) that they make some design choices to prioritize sonic (audible) performance even at the expense of measurements. So it sounds like Gustard are embracing the ASR culture more reluctantly, whereas Topping are all in. Again, I fully understand those who would say that there can be no conflict, because measurements reflect everything we can hear. I don't mean to stir up that debate here, merely point out that the focus on measurements affects design choices of manufacturers in different ways.
Edit2: Oh, what I am thinking: Starting with X16 Gustard IS all in with the measurement culture, providing some of the best measurements of all DACs. Even more so on their newly announced X26 Pro. And it's working, X16s are flying off the shelves. I really wish somebody would post listening comparisons of X16 vs A18. I asked one member at ASR who owns both, but it's not the right place for such a question.
Thank you for continuing to share your experiences with the A18. You're sure putting that DAC through its paces. That's excellent!
About the DSD limitation you found with Audirvana, that makes sense. I saw PCM and DSD limitations too when testing Volumio on RPi, but none of them relevant to real-world usage. Anything above PCM 192 KHz or DSD256 really is just for testing, not practical delivery formats for music. Even among those those want to upsample in software I've seen comments that DSD512 actually hurt more than it helped -- in addition to using obscene amounts of CPU (if upsampling on the fly) or disk space (if stored).
I was glad to see that you found I2S on the A18 to be pin compatible with your Cayin N6ii, as I understand it isn't standardized and can otherwise require custom cables.
But I am VERY surprised that you are hearing differences between USB and I2S on the same DAC, since both are feeding a digital stream to the same DAC chip and playing through the same output stage. The explanations I can think of are:
1) Difference in jitter from the different input circuitry and/or phase shift caused by I2S using an embedded clock signal from the source. Amir on ASR measured I2S on a Gustard X26 two years ago, to dispel the myth that I2S was inherently superior. He found significant phase shift in one of the channels on I2S. It could be just the units he tested or whatever, but it's definitely not a good thing, even if it isn't audible. (Let's not get into the war between extremists of ASR and head-fi. That's not my point here. I believe in both.)
2) Difference in the digital source: If one (say USB) is bit-perfect (being upsampled by the DAC chip) and the other (say I2S) is being upsampled by the source (in this case by the Cayin) then that could definitely cause a difference, as they are likely using different upsampling algorithms. I would be interested if you can still hear a difference if both inputs are fed 16/44.
3) Different MQA processing: I am surprised if the Cayin can do a full unfold of MQA and pass it digitally to the Gustard. I would have thought it could only perform a first unfold and pass 88.2/96KHz when working as a decoder. If so, the I2S feed would only have a partial unfold by the Cayin and no MQA filter applied. (The A18 can only process MQA through USB, as you know). Some people will argue that the less MQA the better, but that's besides the points for this discussion, as there could (or rather: should) be a difference between USB and I2S caused by this when playing MQA. Even if the Cayin can perform a full unfold (say to 352.8KHz as shown on your display) there still would be no MQA filter used by the Gustard when fed through I2S.
As for the difference between Topping D90 and Gustard A18, I much appreciate your posting about this.
I am skeptical that D90's XU208 vs A18's XU216 would make an audible difference, since (again) it's a digital signal, which presumably arrives intact into the same AK4499 in both DACs. But I guess slight differences in jitter are possible.
The output stages are far more likely to cause a difference. AK4499 outputs an insane amount of current (arguably too much) which has to be tamed and converted into voltage (I/V) in the output stage. Not a trivial task. D90 gets it down to 4Vrms (and provides an XLR/RCA switch) whereas A18 lets through a full 6.3Vrms (and no switch). And still, D90 has better THD+N, which makes me think Topping simply spent a bit more on their output stage.
But that's all speculation. I am no electronics expert anyway
Edit: OTOH, Topping was the first vendor to focus squarely on measurements, essentially using ASR as primary channel to promote their gear. And they have been rewarded with valuable recommendations from Amir of their affordable, yet well measuring products. Conversely, I remember seeing quotes from Gustard (sorry, I forget where) that they make some design choices to prioritize sonic (audible) performance even at the expense of measurements. So it sounds like Gustard are embracing the ASR culture more reluctantly, whereas Topping are all in. Again, I fully understand those who would say that there can be no conflict, because measurements reflect everything we can hear. I don't mean to stir up that debate here, merely point out that the focus on measurements affects design choices of manufacturers in different ways.
Edit2: Oh, what I am thinking: Starting with X16 Gustard IS all in with the measurement culture, providing some of the best measurements of all DACs. Even more so on their newly announced X26 Pro. And it's working, X16s are flying off the shelves. I really wish somebody would post listening comparisons of X16 vs A18. I asked one member at ASR who owns both, but it's not the right place for such a question.
Last edited:
