AGAIN - trying to understand technical differences in Coax vs USB connection

Hi - I am still having difficulties trying to understand where the differences in connection come from, which is superior from a plain technical point of view and why they make a difference in SQ.

Also - YES - I have done my research and reading on the internet and in other threads and I know it comes down to the compatibility of the DAC and the source/streamer.

It appears to me though, that whilst recognized pioneers and manufacturers such as Schiit clearly suggest, that Coax would have to be vastly superior to USB “in general”, the most used connection still is USB.

I have tried both and find it hard to compare, as they seem to sound completely different.

Whilst Coax is more powerful and warm/full/rich in sound, USB tends to be a lot clearer and more airy sound, lacking the fullness/richness of Coax though.

I am mainly asking because I am wondering if this is mainly due to differences in the quality of the interconnects used and if it`s worth diving deeper into the differences by getting equal cables (and will the differences vanish when using same quality cables) or do the differences really result from totally different technical aspects when sending the signal via USB compared to Coax.

From what I understand it also has to do with the clocking/reclocking - so the USB package has a clocking in the bridge and sends a perfect signal to the DAC, whilst when using Coax something needs to be reclocked in the DAC??? - thus making USB the more perfect connection - if so though, why do most manufacturers clearly say Coax is the better way to USB???

Just trying to understand stuff here and find the right solution for me (next to letting my own taste/ears decide - obviously :wink: ).

This article sets it out pretty well, gives a history and includes measurements. All modern USB connections are asynchronous.

This post by Brian is also very helpful in understanding the various clocks involved:

Edit: Corrected bad link.

Read it, tried to understand it, still not clear… things were certainly easier when connecting a CD Transport :wink:

Say u have a streaming bridge with the same clock for asynchronus USB and Coax - both sent to the same DAC… the asynchronus USB would be at advantage as for the Coax the DAC would also need a clock that matches the clock in the bridge?

Is that about right?

And - if so, why do Schiit for example state that Coax is the superior connection but still everyone uses the USB connection?

It’s like asking how the sound quality of a CD differs if it is delivered my UPS vs. FedEx.
Coax and USB are only asynchron tranport systems for packets and both are able the deliver the packets unharmed. If one of the services is better for other reasons as sound quality is another discussion.

1 Like

Ah - now that’s an interesting comparison even I can understand… :wink:

Try this Christoph.

He uses lay people’s terminology, gives measurements and a clear winner.

1 Like

Perhaps the most important line …

Does it sound different?
No, none of these measurements indicate any audible difference. The measurements where different, show really small levels of noise and distortion. I can’t make a case for audibility of any of it.

1 Like

… well - it looks like the “experts” can`t really settle on an “only truth” when it comes to technical advantages/disadvantages but the general consensus seems to be, that Coax and USB basically should not make any audible difference either way… thus - FREE CHOICE :wink:

Yes, but … my original endpoint was a Raspberry Pi and the USB implemented is far from brilliant. I could hear audible ticks and clicks on my Chord 2Qute, so opted for the Allo DigiOne hat (S/PDIF.) However, if I had a CuBox-i4 I think USB would’ve been perfectly fine.

I think that might have more to do with interaction between USB ports and network interface on the R-PI. Some DACs seem very sensitive to this and some not. My dragonfly red seems very robust for eg compared to my pro-ject S2 digital which seems very sensitive (better since disabling the ethernet) - despite both playing the same album via Roon.

Re topic - USB vs SPDIF - depends very much on implementation of the various bits involved. SPDIF is simpler, so less to go wrong, OTOH the DAC has less freedom and no help with how it maintains sync.

Most USB issues tend to be down to inadequate software implementation either in computer driver end or in the DAC device end. The days will long stable class drivers in the computer, then non-ideal implementation at the DAC end might be more to blame (cough pro-ject/JW cough).

With USB there are two modes - isochronous and asynchronous. Isochronous is in its basic form very similar to SPDIF - send the data and hope and let error correction/detection deal with any issues, and let the DAC sort out adaptation/sync. It uses framing and so jitter is not an issue, unless the data rate issuchn that the DAC is struggling to maintain sync. At the other extreme is async where the data is fed according to the demand of the DAC in blocks so is exactly at the right rate on average. Isochronous can be used in a hybrid manner where there is feedback to tweak the sending rate to perfectly match the DAC clock (ideal). Over a quality connection, then this is probably an ideal - low latency with a data rate on average that exactly matches the needs of the DAC.

And then you have DACs where there are a variety of strategies to deal with clock extraction and synchronisation. The ideal is always that the data arrives at exactly the right rate on average for the highly accurate internal DAC clock to be used with no need for external sync - this is USB (isochronous with feedback or async) only. A DAC may choose to synchronize with external clock, or it may choose to resample (perhaps combined with upsampling) or some hybrid. Some DACs do the conversion of the original sample rate, some run internally at their maximum rate. At 768K with 32bit or better internal processing and high quality algorythms, you shouldn’t really care how it goes about it - the result should in theory be good regardless. Sometimes the algorythms are simply not very good in which case it is best to avoid them entirely if possible (which USB may allow for, but SPDIF will not).

If you can use USB and the result is clean - then use it. If th8ere is something wrong with USB reliability, then use SPDIF. That apparent punchiness some people describe with SPDIF (?!?) may well be the application of a poor resync strategy, or poor interpolation in resampling, or poor dithering choice somewhere. Try cranking the level a bit - does that ‘punch’ start to get harsh in mid/upper instead? If so, then well done - you spotted a poor conversion job - time to find a suitable forum to go rant about it :wink: It if doesn’t get harsh, well maybe you have a very expensive AKM chip based DAC which IMHO do tend to be quite punchy anyway :wink:

1 Like

That’s correct, LAN and USB share the same bus on the RPi. My point here is that one connection type may prove better than another depending on the kit you already have. It’s worthwhile asking forum members who use the same DAC to share their experiences with different endpoints (Roon bridge.)