What effect does OS or hardware have if you use a decent, reclocking DAC (so that input data jitter and noise don't affect your DAC master clock)?
The process from the computer CPU, till the I2S data reaches the DAC chip is very complex. There's a lot going on that affects the sound. As far as operating systems go, Jussi Lakko from HQplayer and Brian Luczkiewicz from Roon are 2 great guys to talk about the issues with OSX and Windows for audio. Both of their audio players sound way better running in command line Linux.
However they prefer to avoid the topic as when people have a bias towards certain hardware and OS's, they simply won't buy your product if you tell them it doesn't sound as good on what they own. The truth is often bad for business.
As far as USB interfaces go, the noise creeps through most interfaces. It can easily be heard. That extra flip flop reclocker after the USB interface in the Phison PD2 is the best method I've seen to eliminate this. The best DAC's slave the DAC chip to the same clock the USB interface uses. Noise from the computer gets into the USB interface chips and adds a pile of jitter to even clock # 2 of the interface. Even if the interface is powered by its own supply. This is why this flip flop is so effective. It's after all of that. No matter how good of clocks you use, if noise gets in them the jitter performance goes out the window. Clocks are extremely sensitive. Even blowing a slight breeze over 1 causes jitter spikes. My clock guy who has a $100000 machine for measuring clock jitter told me this.
Last edited:
I remain open minded and would readily consider alternatives is the manufacturer can demonstrate the improvements they offer. My DAC has the M2tech hiFaceTWO USB3 interface.
I plan to do measurements. I'm not sure how that USB interface is built. Hopefully much better than the M2tech one I have right now.
The level of improvement you get by upgrading the transport all depends on the USB interface implementation. With the M2Tech OEM I have here the difference was night and day. Same with the Amanero. Galvanic isolation like the XMOS based DIYINHK DXIO has helps, but they also add jitter in the process. I find with a dirty computer the DIYINHK interface sounds better, but with clean streamer the Amanero sounds better.
Have you tested any interfqaces/DACs that incorporate bothe galvanic isolation and buffered reclocking?
Not really. The only two factors are jitter and noise.
Can you explain what the effect of the reclocker is in removing noise (reclocking is normally used to remove the effect of jitter, not noise)?
Usually it is cheap and simple DACs that do that. More advanced ones isolate the clock domains.
Wind is not the same as noise, and the one with the most expensive machine is not necessarily right... 🙂
Sonny explained it better. Noise causes jitter, so when you reduce the noise, you reduce the jitter.
Galvanic isolation on its own isn't enough. It reduces noise, but adds its own jitter at the same time. It took Gordon Ranken years to realize that his perfect design was highly flawed. But he realizes it now. Unfortunately many of his followers are still back in 2009, while he's moved forward.
Yes expensive machines are required if you want to see real world results. You can see the phase noise spike in the screen from vibrations and wind.
And do you hear a difference in the sound with different sources even with those?
Noise causes jitter only when the noise has a way to affect the master clock.
So if you want to be 100% sure of no interference, you buffer and reclock after the galvanic isolation.
But are those spikes audible?
Much less. Although switching to Linux on any hardware still sounds much better. Especially with the XMOS chips. If you want the best sound out of them, feed them native DSD only. DoP causes audible degradation. Although it's a lossless process in theory, the additional processing causes the processor to work harder in the USB interface. It can easily be heard.
I suppose whether it's audible or not depends on the hearing of the listener. A tree falling in the forest is inaudible to a deaf man.
Really looking forward to measurements showing those differences. I assume that by the same logic, you will say that a WAV and FLAC file sounds different.
Yes this is what I'll be trying to do with the aPX-555. It looks better to people who can't make it for a live demo to see an improvement on paper.
However so far anyone who I've actually demoed this stuff too agrees with my findings 100%. When I demo, I never let anyone know what I've changed. 100% of the time they pickup on the same things I do. But my speakers and amps in my upstairs casual listening room are extremely revealing. The Focal W cone mids run full range with only Neotech UPOCC copper wire between them and the NC-500's. And the Raven ribbon tweeters only have Mundorf Silver/oil caps, copper foil inductors and Dueland CAST resistors in the signal path.
Not only that, but many of your claims are rather bold and go against many of the measurements and tests other people have done, so verifying them with measurements would be good, especially as the measurements are easy to do.
I hope you realize that you are describing a single-blind, not a double-blind test. I assume you have heard of the Clever Hans effect?
ExaU2I uses an FPGA to buffer I2S stream and read out with high quality clocks. But then the isolator is plonked right on the output I2S interface.
Any guesses on why that choice was made?
I don't really care. I learn by hands on experience. Not blinding parroting what others say.
My test methods are perfectly fine for my needs. They result in sales every time, as well as years of satisfaction for my clients. As I said the only reason to post measurements is for marketing purposes for online sales.