two hours in summer term, and then gave up
Not to presume of anything, but on top of the great stuff that’s been suggested, you might want to make sure expectation bias isn’t polluting what you’re hearing too, too much… let’s start by not overestimating the human mind, shall we
Bear with me one second:
So in my study/listening room, I have a computer monitor hooked up using an HDMI to VGA adapter. I noticed that the monitor seemed to blink when I’d walk by too closely to it.
I then realized that the monitor would blink every time I touched the desk/rack that the monitor is sitting on. The desk is a metal frame with wood shelves. I could touch the desk ANYwhere - leg on the other side, other wood piece on a different shelf, ANYwhere. That would make the monitor blink.
Static electricity. I changed out the adapter and the problem stopped. But the point is, if you can touch a piece of wood 4 feet away from an electronic device and with an imperceptible static charge affect that device, then tiny spikes and such from a power supply over USB from a computer sure as heck can affect some aspect of digital to analog processing or the handling of the signal on the analog end.
Isiah, try this - find a USB DAC that doesn’t draw power over USB. See if those two computer sources sound as different then…
The other thing to remember is that no matter how the data gets to this point, it is inherent in the digital to analog conversion process that the data be in effect streamed…it is not the same type of logical process that goes on in a CPU or that transfers data in packets across a network, two processes that are much more resistant to errors. The D/A process is much more sensitive and prone to errors. These tiny errors can affect sound quality. So an electrically quieter USB source can improve sound quality, IMHO.
There are computers (motherboards actually) that have USB ports which can have the power feed disabled. Check out the “DAC-UP” ports on the Gigabyte motherboards.
But does it? Archimago tested the proposition by measuring the results and found that it did not:
He measured what he could measure with the tools and knowledge he has. Given how much we don’t know still about human hearing, there’s little reason to believe that the standard measurements are enough to capture subtler perceptual distinctions. Standard measurements are enough to separate junk or faulty gear from well-designed and well-functioning gear, but that’s not what we are discussing here.
But that way lies madness! You can’t prove a negative: “Maybe there are ghosts… After all, no one has proven there aren’t!” He measures what we do know about human perception, using the best current model of human auditory processes. Any subtler perceptual distinctions must be so subtle that we haven’t been able to see/hear them yet.
Not to say there aren’t electrical effects, particularly for DACs that are USB-powered from noisy computers (which he didn’t test, to my knowledge), and for DACs with crumby input stages or otherwise indifferent engineering.
Maybe. In some cases. In USB-powered DACs with poor power conditioning, or otherwise poorly engineered electronics. But as a general rule?
You know, while Archimago’s stuff, and Amir’s tests, and Atkinson’s stuff in Stereophile is nice, there’s a certain something missing here. Audiophiles fret about power supplies and power conditioning. You buy a $5000 piece of audio hardware – do you now need to buy another $2000 power conditioner to keep it sounding good? Wouldn’t confirming or allaying that fear be a great service to provide to the community?
Why don’t these objective testers (and Stereophile) test that part of the equation? Why don’t they run power sources of known quantified noisy-ness into these devices and test how well the device’s power conditioning works? Why don’t they send USB signals into USB-powered devices with known quantified noise on the power lines of the USB cable, to see what happens? Why don’t they test with and without super grounding boxes to see if they are necessary?
I’d expect that manufacturers follow some variant of the 80/20 rule – they provide power supplies that are good enough to handle 80% of their customers’ mains power (or USB power) conditions. Maybe it’s 70% or 90%, but something like that. But it would be great to have someone who gets their kicks out of testing to be testing that, and let everyone else know.
Absolutely not. All of his measurements are based on time averages with periodic or impulse inputs. That’s all very nice if linear systems theory and Fourier decomposition is all that matters, but human hearing does not conform to those old models. There’s a lot of hearing psychophysics since the 40s showing that human hearing does things that the older models cannot capture (see the book I linked to). I used to hang out with the people at Bell Labs working on audio compression (co-creators of perceptual audio coding with Dolby). I learned then is the old “hi fi” measurements were sorely lacking when trying to predict human response to signal processing. On one hand, masking phenomena allowed effective compression that threw away parts of the signal in ways that would show as high distortion under the old metrics; on the other, sometimes those perceptually effective compression methods failed miserably on human evals even when they measured well.
But the question, I guess, is not about hearing. My mistake in framing it that way. (By the way, that’s a large book; are there particular chapters you’d like us to look at?)
It’s about fidelity of reproduction. Does the output signal of the DAC faithfully reproduce the signal encoded into the digital file? That’s the question to be answered, and that’s what those testers like Archimago look at. What signal needs to be captured, and what the best way to present that signal to a particular human listener, those are different questions, and bound up with hearing. That’s where filters and distortion and tone controls come in. Sure, you can fiddle with that in the digital domain, too, but that’s just producing another signal which you’d like your DAC to faithfully image.
And yes, I agree that Archimago and his ilk confine their tests to regions of the signal spectrum which are conventionally thought to affect human auditory perception. The conventional Redbook capture domain. Perhaps more should be captured. Perhaps higher-res recordings actually do capture effects beyond that domain. However, I doubt that any of those possibilities would explain why two different streamers taking the same bits to the same DAC would exhibit different sounds.
Very interesting sharing! Like it so much.
I do have that kind of experience but not so dramatic. My Amp, subwoofer would hear pop sound from turning on other equipments. The problem will be gone if I plug them into a powerbar.
Talking about noise, I was very skeptical about the significance about it to modern digital equipment. However, recently I re-plugged my Sony Bravia TV from a cheap power extension broad with two NAS (each got fans and 4 hard disks) to the TV outlet of " Belkin AV Isolator Home Cinema Surge Protector"
The video became more vivid and clean. (raised by my wife and she didn’t know I changed something). So I agree that noise can sneak in and affect the video or audio quality.
Another experience is if I connect my amp directly to my Roon Core, the audio output is full of significant noise. The case is much worse than using Mac or dietpi as endpoint.
Back to my topic, no, my USB DAC doesn’t draw power from my Mac. Actually the manufacturer said that the USB power pin was disabled. Honestly I think my Mac is quite quiet as there’s only SSD. The cpu fan is in quiet mode most of the time.
Motors (as in fans and hard disks) are notoriously electrically noisy. I would guess your NAS’s were leaking that noise back into the mains supply. And your TV wasn’t filtering it out again.
So far I didn’t see Archimago comparing the result sound wave of a same song from different setup, or I don’t understand his diagrams?
I mean I tried to connect my USB DAC to the NAS Roon Core. Very significant noise on the music. Not just change of sound signature likes Mac Vs diet pi this time.
“Faithfully” according to what measurements? All the measurements used in these tests are time averages of simplistic signals. Their supposed adequacy is based on simplistic theories of sound reproduction and perception. They are enough to detect gross distortions, but not to capture all aspects of the output waveform relevant to how it might be perceived by listeners.
I think we can suggest Archmiago to measure the full wave form next time.
Yes some DAC can still sound different even if the 5V line is cut.
Always fun to read topics like this.
I’ve always wondered why everyone - from audiophiles over audiophools and snake oil salesmen to people with severe cases of FOMO - spend so much time pondering what might or might not interfere with a digital signal. If indeed anything does.
I wonder why no one ever pays attention to the stage where the signal is most likely to suffer unwanted changes: once it goes analog. Amps and speakers.
Something to ponder: with many amps, no two analog inputs are truly created equally…
Well, that’s a plausible idea, anyway. Not sure I believe it. Do you have a description of what an adequate testing regime would look like? Or can you cite chapter and verse from Lyon’s book that pinpoints what you think is missing? I downloaded the author’s draft and have skimmed it, and have to confess I don’t see what you’re driving at with your reference to it.