Ok, so why would I want to use USB at all? Before you were telling us that USB is better than AES/SPDIF, but then you just listed all the problems inherent to USB that dedicated audio transmission protocols don't have. Something doesn't add up.

BTW I know some guys that would laugh at you for saying Firewire is a better transfer protocol for audio. Where are you getting this from?

Until a few years ago, you would not want to use USB. If you remember back to the DACs that first started appearing with TI PCM2704 based USB inputs, even the manufacturers themselves would say look, the input is there for convenience. It's garbage and we don't really recommend you use it unless you have no other choice. That was then.

The PCM2704 was replaced with newer, better performing chips, which were in turn replaced with newer, better performing chips. I don't think Wavelength "invented" asynchronous mode streaming to replace the old adaptive mode, but they did popularize it. Asynchronous was a whole new ball of wax for USB, making it much closer to what FireWire already had. The latest XMOS chipset and software stack, when properly implemented, can be very good, which is likely why FireWire has largely died out except on professional multi-channel AD/DA products.

So with the input out of the way, you then have to look at the output. You don't want to use a standard motherboard port. It's incredibly noisy, and everything basically ends up in a big bucket along with USB keyboards and whatever other devices you have attached. What you want is a dedicated PCI-e USB card, with a dedicated power supply. The SoTM card was one of the first, and they have a newer model with a USB 3.0 port, but the JCAT card I think is the current gold standard.

Once both the input and output are up to snuff, USB will outperform AES/EBU outputs from Lynx and RME sound cards. It will also outperform most lower cost Ethernet based network streamers, as well as typical disc spinning transports (IME). I don't know how something like the Aurender W20's USB output compares to its AES or S/Pdif output, or how it compares to something like a Naim NDS streaming via Ethernet as opposed to pulling from local storage like the W20.

USB will outperform AES/EBU

Could you explain this?

But most DACs won't clock off the digital inputs but rather use their own interal clocking/PLL to generate the clock. It's asynchronous just like USB. I feel like that argument falls flat on its face. Also nice how AES/SPDIF is transformer-coupled in good implementations so you're free from any noisy ground.

Out of curiosity, do you think USB would beat a dedicated transport with AES/SPDIF out? I've yet to experience a case where USB would beat that (or AES/SPDIF in any scenario really), but maybe I need to experience special JCAT card.

Indeed, this is also my finding. There a few ways I heard to noticeably improve USB sound, some stacked or separate:
- SOTM network streamer with LPSU (my preferred, since no PC at all is the best PC from my experience)
- SOTM USB card with LPSU
- using Y-Split USB cable connected to a 5V LPSU. It will clean-up the power line, but this won't help the data line noise, which will get to the DAC.

Other stuff:
- Schiit Wyrd
- SOTM Battery PSU for network streamer
- power the LPSUs with a mains power regen
And a few more I probably forgot.

I am still looking to try a couple of CD transports people recommend here, to hear how they fare. I didn't get any other suggestions beside Theta Data 3 from the other thread. The key is resolution - it has to resolve to even properly compete.

" LOLLOLOLOLOL" is all Ls and Os! LOL.

I am in in the all 1s and 0s camp, but I hear what you say about USB. With USB, it is not so much the 1s and 0s that bother me, it what might go along with them.

Yes, USB would be much better if it wasn't otherwise sending 5V of power along with the data stream. People have reported significant improvements by simply putting a piece of tape to block the power leg pins for DACs that don't need bus power.