How do you measure this? With a 1 kHz sine wave? And then if the signal looks nice on a scope it is an exact copy?
Or did you check the complete space of say 44100 samples multiplied with 20 bits and tried all possible stimuli within this space for equivalence? Or do you assume that with a few sine waves the complete DAC transfer function is characterized?
There are many industry standard test signals, and you can also do a nulling test with your favorite music, but I assume you are familiar with the Fourier series?
To my understanding, an ASRC still has a PLL to estimate input clock rate. So from that perspective there is still some uncertainty on the input clock phase. Apart from that, there is little value of a jitter less digital domain PCM word. In the end it needs to go out of the ASRC chip and become a digital signal in analog electrical domain. Even if all jitter would be removed in between, supply, ground-plane and EMI interference would bring back some jitter.
(This is also the case in asynchronous USB solutions. The amount of "effort" the XMOS chip needs to do to recover input data has supposedly a measurable influence on the supplies which causes jitter on output side.)
On top of this, the question is what degradations the ASRC interpolation would give. It could be that the level of these degradations is jitter dependent.
Yes, that is the typical ASR vibe... ;-)So being evidence-based marks someone as one of those horrible ASR:ers?
I believe DACs can sound different and there are people who can hear more and finer of these difference than most people. So "statistically" doesn't apply to all people. But such good hearing is so rare, and everyone cannot have that. Otherwise they would not have statistically "better" hearing.
I have never heard these Hypex FusionAmps through their own AD converter, I've always driven them with a digital signal. I do hear hear the hiss they produce through my horn loaded compression drivers (= they are very sensitive), and if I were to feed them analog signal, there would be more hiss. I do not know exactly how much so or if I could tell the difference, since I almost need to put my ear on the mouth to hear it with digital feed anyway.
There are things like jitter and other problems of magnitude of an earthquake (sarcasm), but how much it really matters? Like sound from a phone charger is louder than jitter. I'm sure you can't go wrong with either analog or digital feed. You need not worry about problems that are not problems.
I have never heard these Hypex FusionAmps through their own AD converter, I've always driven them with a digital signal. I do hear hear the hiss they produce through my horn loaded compression drivers (= they are very sensitive), and if I were to feed them analog signal, there would be more hiss. I do not know exactly how much so or if I could tell the difference, since I almost need to put my ear on the mouth to hear it with digital feed anyway.
There are things like jitter and other problems of magnitude of an earthquake (sarcasm), but how much it really matters? Like sound from a phone charger is louder than jitter. I'm sure you can't go wrong with either analog or digital feed. You need not worry about problems that are not problems.
Indeed, aren't those series calculated for an infinite amount of time... (and basically assume zero jitter and other degradations)... ;-)
There are easy ways to account for non-infinite time (otherwise you couldn't do spectrum analysis in finite time), and jitter and other degradations show up as (measurable) noise and distortion.
For the record, I am definitely not anti-science. Actually quite the opposite. Still or especially, I like to keep an open and exploratory mind & try to find (scientific/technical) explanations for observations done.
But I do not like to up front exclude the possibility of observations on basis of previous research or the quite limited audio measuring methods & related assumptions that are there. Like that a SINAD above the cooking temperature will result in a perfect sounding DAC...
But I do not like to up front exclude the possibility of observations on basis of previous research or the quite limited audio measuring methods & related assumptions that are there. Like that a SINAD above the cooking temperature will result in a perfect sounding DAC...
According to this ASR backed science you seem to love so much, any live instruments should sound much worse than through audio system in an anechoic chamber.That is what the anti-science types seem to think.
Would you mind sharing your thoughts on controlling the volume in this setup. NCores seem to like a very strong signal but PC is likely to have a better digital volume in you use players like Jriver. To be clear I don't know how good FA mechanism is but some comments in this thread mentioned better sound quality when adding gain instead of lowering it.
That is so far off that it isn't even a proper straw person. More like a haystack.
Just get an RME audio interface. They are really good in terms of jitter. (and most other stuff too ;-))
For the record (whatever record it is) I like too keep an open and exploratory mind too, but also question my biases, perceptions and assumptions - that is the basis for science. Thus I also take into account the properties of the human mind and perceptory system, and realize that rather often, the proper explanations are not technical, and belong in the science of psychology and perception.
It is not a question of "liking" a signal, but of course a proper gain structure is important if you want to optimize signal to noise ratio. Ideally the gain allows the amp to produce the maximum listening level you want with a full amplitude signal and volume turned full up.
I actually have a RME AIO PCI express audio interface. I did not recently use it for S/P-DIF, but I do not recall it was so good jitter wise...
Still I like RME a lot, mostly for good & long term driver support. I have a nice pile of internal and external soundcards in the attic that are worthless due to lack of driver support... :-(
This is a good question. I think I already shared some words on this earlier in this thread, but let's provide a full summary here.
First, from my experience with a XMOS based DDC & PC it is jitter/audio quality wise best to transfer in native (lowest) bit depth. So 16 bit for red book CD material, 24 bit for HD stuff. Doing some volume control on top of that would result in deterioration. Only reason to do upfront (PC/streamer side) digital attenuation in a Fusion setup would be to avoid intersample clipping in ASR. Not sure whether this really occurs or not. I did some listening experiments with 1-2 dB and ~6 dB (exactly factor 2, 1 bit) attenuation (at 24 bit bit depth), but cannot recall exactly the results. I thought in most cases I would prefer 0 dB driving (and 16 bit for red book).
On Fusion side, I have the impression it is beneficial to apply maximum channel volume (if possible). I have all channels on max volume (15.x dB). If I need more gain, I do with an additional filter. Attenuation I do with master volume (via remote control).
I maximally reduced Fusion Class D pre-amp gain by modification (removing some resistors, search back in the thread). I do not need the headroom with my relatively sensitive drivers. On top of this, I put series resistors on mid and AMT drivers. Mundorf M-resist Supreme. This way I think have a reasonably optimal gain structure for my speakers. And the series resistors also provide measurable frequency response benefits (and likely somewhat lower distortion as well).
I can share mine. If you feed digital input, it almost don't matter one bit (pun intended) how you control the volume as long as your gain structure reasonable. Digital volume control is a multiplication arithmetic operation and it's very easy to get it right.
Lowering volume raises relative noise floor, but if you use digital input, the noise floor is as low as it theoretically can be. And the bits that you loose, are the quietest ones that you wouldn't hear anyway. If you are afraid of losing bits, the only option is to play back at nothing but full volume. Analog attenuation introduces more noise than digital, because analog components add noise among other problems.
I often feed my FusionAmps a digital signal that is attenuated by 50 dB, that's for when the kids are sleeping and I'm still watching TV. I tried to lower it to -60 dB and that's closing to a level where the sound is barely audible at all in the listening spot. Even then I have not noticed any audible defects in the sound coming out of my speakers. The DSP is able to process and cross it over between elements without any trouble. -60 dB equals about 10 bits of attenuation.
If you read ASR, be sure to read the whole post containing the measurements and not stop after SINAD. With comments like that you almost sound like you think that SINAD is the only measurement in the world.
And what comes to audio science, it might be true that we or they don't know everything about the hearing and don't have all the equipment in the world for tests. But it's stupid to think that you need perfect set of everything before you can stay doing anything. The fact is, we do know heck of a lot about sound reproduction and how to measure it. The data on ASR is scientifically much more valuable than random nicks posting comments on a discussion board or magazines for that matter, sometimes speaking in wine terms. I hope you don't fail to see that.