USB Nervosa Thread Decrapifiers, pro interfaces, and bears oh my

Often, even if the DAC is externally powered, the receiver isn't. Moreover, even when the receiver is powered from elsewhere, it often uses that VBUS power line to sense the connection. You almost certainly need power on it, though that doesn't have to come from the computer itself- the easiest way to achieve this is a Schiit Wyrd, which will give you cleaner power from an LPS on that line.

(Don't spend much more than that on cleaning up USB, though- use a different connection or put the money towards a better DAC that works with different connections.)

 

Thanks Kattefjaes

 

Update from @schiit on HF today:

"With respect to USB/SPDIF nervosa:

1. It's important to keep in mind that most of the changes we're talking about (when discussing USB decrapifiers, converters, etc) are relatively small, at least IMO. I use 'naked' USB all the time--yes, just a cable, sometimes not even a fancy PYST cable, and I feel absolutely no audiophila nervosa about it.

2. It's best to get a great DAC and see if you like it without any decrapification/conversion. Who knows, you may be like me.

3. If, after getting a great DAC, you decide you absolutely, positively have to try some kind of decrapifier/converter, it's probably best to wait a bit before spending tons of money...just saying..."

 

Yeah don't go crazy with decrapifiers etc. Fix the problem at the source. That will either be the DAC, digital inputs, bad or excessively long digital cable, or the computer. To be honest who knows what else - it seems upgrades are limitless but nothing loses value faster than digital tweaks.

 

I know I remain the odd man out, my views aren't going to gain me any plus points, but I still see a huge difference between the needs of a real-time playback system, and a play back of music recorded in the past.

In a real-time playback system streaming digital values, an accurate clock matters, a lot.

For playing back music recorded in the past? The original clock is long gone, and I still see no reason why it matters that the USB packets are clocked perfectly. They are packets, and the packet clock has nothing to do with the code-word clock. So long as the receiver (i.e., the DAC) is not starved for data, who cares what the packet clock is? The DAC can't possibly be using that to clock out codes into the pipeline, or if it is, what a lame design.

I mean sure it's possible that the USB implementation is screwy or adds noise at the analog level, but RAM is cheap, buffer that data, read ahead, clock out the data at the original recording clock (i.e., to the known recording rate). If the problem is the original recording clock is long gone, I can't see how it matters if you spew that data over SPDIF or USB packets, or Ethernet packets. It's gone.

On the flip side, USB is not inherently isolated. There are boxes that provide this, so no ground connection from the computer to the DAC, but still has nothing to do with the USB packet clock rate.

I also wonder how much of this belief USB sucks is due to people taking their 16/44.1 music, and up-sampling to 24 (irrelevant)/96 (relevant) vs say 24/88.2 (2x, an integer multiple of 44.1).

 

It has never been argued that clocks and data delivery are the problem with modern USB - Torq has posted enough times on the subject about the workings for Dunning-Kruger afflicted enthusiasts - but that, at bare minimum, for subjective reasons, AES EBU is the preferred interface choice. Hydrogen audio serves as the best outlet for those that want to spew technical details without having implemented a solution before.

 

@Darren G it's not some sort of myth. Out of the available options, USB sounds the worst, and that will be true for the majority of systems. Whether you hear that or value that change for the cost it takes to implement non-USB solutions is up to each individual. For me, I will never listen to USB again. The difference in my system is not at all subtle.

 

Recently I've been going through all of the albums I have listening blindly and rating them 1-5 in Roon based on how good they sound (subjectively). I was surprised that about one third of the albums I've rated as 5 were mp3-s, where I rated lots of hi-res or flac converted from dsd only as 3 or 4.

The other day I was playing with Jriver and Sonnarworks using the usb out and I forgot to change it back to the AES input I usually use.
The next day I couldn't understand why everything sounded bad. Went back to some albums I'd rated as 5 before and they sounded worse too. Then I relized what the problem was, changed back to AES and everything was good again. Then I played a little bit with a few different setup using USB, coax, BNC and USB definately sounded worse than the other connections. Very noticeably worse, in some cases almost like a different mastering.
It might depend on the system and the implementation of the conections but on my system I'm sure I will avoid USB.

 

There isn't any difference in music recorded yesterday and 50 years ago with regard to clocks. The reason people talk about them is they're a necessary step to decode the bitstream and the better / more accurate the clock you're using the more accurate the reconstruction will be:

https://en.m.wikipedia.org/wiki/Pulse-code_modulation

There are a lot of theories about why USB is bad and I agree that it seems that you could avoid at least some of them by making the DAC smarter but the more complexity you add to that box the more likelihood your solution causes some other problem (e.g. electrical interference). You could try the raspberry pi + digital out hat solution that some have done here which accomplishes a lot of what making the dac smarter would and is relatively inexpensive.

 

For Yggdrasil and Gungnir Multibit, Schiit indicates they are re-clocking anyway (presumably at least some other DACs do so). Makes sense to use a DAC side clock vs depending on a clock in the data delivery stream being delivered jitter free (that's for streams; for packets, depending on packet arrival time is going to be error prone to have any dependencies on).

I sometimes wonder if the Adapticlock feature is a key reason they sound so good? Could also be their Adapticlock works best with non USB sources, though the sound for me has been very close on Yggdrasil (could be the Gen 3 USB)?

 

Any modern USB is async and has to reclock.

 

Yep, so that leaves -

The bits are wrong over USB (unlikely).

The GND adding a ground looop, or VCC line introducing noise.

The re-clocking isn't so great over USB (possible I suppose, if say re-clocking the code words is not purely async, such as an interrupt to handle an incoming packet or other processing).

The USB implementation introducing noise into the chassis and circuits that the other inputs don't.

Packet re-clocking still seems pointless, but oh well.

 

Good shot over the bow, but I still disagree, and hey if I didn't, maybe I'd be Jude vs Tyll (damn him for not joining the Sony rocks no matter). Could be way wrong, or could be mass belief at play too. Time will tell

 

Trying to save you time and money here. Feel free to do whatever pleases you, of course.

 

So the never ending battle of USB or Not rages on,

what is a good transport to switch from "dirty usb" to a cleaner SPDIF output, what should I start with. My aim is by January to upgrade my Dac/Solid State amp and re arrange my work space so I have room for more hardware. Price wise I'd like to stay around $1500 for a dac solution

 

Wait for some new Schiit to be released.

 

Auralic Aries. Raspberry Pi with Digi+. Mutec. Lynx.

All mentioned already in this thread.

 

As others have said, I would wait to see what Schiit has coming out. But, if you were in a rush, the Singxer F-1 has been fantastic with my Yggdrasil (and Modi Multibit) and is inexpensive (~$200).

 

Thank you for that! I saw it on amazon and figured that'd be the way to go. Wired that they don't sell it with the shell but, no bother I can slap it together