USB vs Toslink feed to DAC - a personal perspective

BMFDrums

New member
Dec 15, 2012
8
0
0
I've seen a number of posts on this topic asking which is 'best' - USB or Toslink for connecting to a DAC. As ever, 'best' is not always easy to define, and "one man's meat is another's poison" and all that. I've been experimenting with kit and connections recently, and have the following to offer.

First, a very Noddy introduction to the two. From a data transfer pespective, each of these approaches has a distinct advantage and disadvantage: Toslink provides electrical isolation of the DAC from the source, but the clock controlling the data is in the 'push' device, normally a relatively poor quality device (leading to higher levels of 'jitter' when compared with the clock in DAC); asynchronous USB transfers the control of data flow to the (normally much better) DAC clock (reducing levels of jitter), but the two devices are (normally) connected electronically, so any electronic interference from the source device can affect the DAC. I say 'normally' for the USB connection, as there are power supplies available that provide the DAC with a separate ultra-linear DC supply that is isolated from the supply to the computer/streamer.

I normally stream music to my DACs (Audiolab 8200CDQ, Beresford Bushmaster, and Audiolab M-DAC) via iTunes and Airplay using Apple Airport Express (AE) units and Toslink optical, sourced using wired ethernet connections. I have 2 x gen 1 AE, and 1 x gen 2 Apple AE - more on that later. Intrigued by the difference that USB connection would make (or not), I recently bought a Raspberry Pi to experiment. Just to put the units in context, the AE are £100 each, whilst the RPi is about £50 by the time you add a power supply, cables, case, and SD Card. Using the excellent Volumio software on the RPi, I was soon able to stream iTunes to a DAC using the USB connection, but the RPi would pop, click, or drop-out every 6-8 seconds, and was too annoying to use. I tried a different (more expensive!) USB cable, and this did seem to make a (very slight) difference, reducing the pops/clicks to once every 9-10 seconds or so. However, there appears to be an issue using wired ethernet and USB out from the RPi, as I explained in a previous post on the Volumio forums:

http://volumio.org/forum/hip-pop-with-rpi-and-usb-dac-t662.html

Whilst the USB connection to the DAC was too annoying for regular use, I did notice that the sound from my main system seemed to be slightly different wth the USB connection (in between the pops and clicks!). 'All' that had changed was that

iMac -> router -> AE -> Toslink -> 8200CDQ -> amps & speakers had become

iMac -> router -> RPi -> USB -> 8200CDQ -> amps & speakers. So, not a massive change in the overall chain from source to DAC, and the amps, speakers, room and set-up remained identical.

Through USB, I thought I noticed that the soundstage seemed wider, and that somehow the music seemed clearer, but it wasn't easy to pinpoint what was different (if anything). I was also very aware that it might just be down to expectation bias, and that there may be no real difference at all, although I couldn't help but feel that it was 'better' via USB. At that time, I had only been using one track that I know extremely well - the live acoustic version of "Hotel California" off The Eagles' "Hell Freezes Over" (ripped using ALAC on iTunes).

It was when the next one of my usual testing tracks came on that I noticed a definite difference, and afterwards swapped back and forth between the AE (Toslink) and RPi (USB) a few times to ensure that what I was hearing was correct. The track is "Limit to Your Love" off James Blake's self-titled first album, and has a massive sub-bass element to the track. Via Toslink through the Apple AE, the sub-bass is very obvious (it's AWESOME!), but the specific notes of the sub-bass were difficult to pick out. Using USB from the RPi, the individual sub-bass notes (i.e. the difference in frequencies) are very clear and obvious. Similarly, the kick drum at the start of the Red Hot Chili Peppers "Dani California" is much clearer and the tone of the drum is very obvious, as are the notes on the the double-bass in Diana Krall's beautifully-recorded "Temptation". In every case, USB gave a clearer and more accurate presentation of the bass notes and (I believe) a slightly warmer and more realistic sound overall. However, that is much more difficult for me to quantify or give clear and definite examples - it is much more a 'sense' and 'feeling' that things were better with USB.

In my testing system (RPi -> USB -> 8200CDQ), I didn't notice any electrical interference or loss of SQ using USB, and I was only using a cheap power supply to the DAC. That said, the 8200CDQ has its own mains power supply, so it is not drawing any current from the power supply to the RPi. If you were to use a DAC that draws power from the supplying USB connection, that would almost certainly benefit from a better power supply than the cheap one that I bought for the RPi. I have also experimented with an on-board DAC for the Raspberry PI, but that is the subject of another story - more via the links below if you're interested.

http://www.whathifi.com/forum/hi-fi/my-raspberry-pi-hifiberry-music-server#comment-2933830

http://volumio.org/forum/rpi-hifiberry-dac-t879.html

I mentioned the Apple Airport Express units above. They are very handy devices, and I'm sure that many people (like me) love the simplicity and true 'plug and play' nature of Apple products. However, the gen 2 AE is MUCH worse in terms of jitter than the gen 1 devices. How do I know? Because the 8200CDQ continuously loses 'lock' with the gen 2 AE, but works just fine with the gen 1 boxes. The M-DAC has a function to alter the tolerance to jitter, and it needs to be set to 'high' for the gen 2 box, but will work on any setting with the gen 1 AE units. I should mention that all of my testing was carried out using the gen 1 box, i.e. the lower jitter device, so if you were to swap from a gen 2 AE via Toslink (assuming that your DAC could deal with the jitter, as the Bushmaster can) to USB, you may get an even bigger difference (improvement). Conversely, if you are connecting your DAC directly to a high-quality computer, the difference that you observe may be much smaller than for me, where the 'weak link' appears to be the AE hanging off the end of my router and house-wide gigabit (CAT-6) ethernet LAN.

Summary. For me, USB is definitely a little bit better than Toslink, mainly because lower frequencies are definitely better reproduced (on my system), and the overall sound is slightly warmer and more natural, although that is much less obvious, and some of it may just be in my head. The whole testing process was really interesting, and has led me to invest in new boxes and kit to ensure that I can get the 'best' supply of digital data into my DACs. I hope that's been useful for people, and helps to inform their streamer/DAC choices.

Enjoy the music! ;)
 

unsleepable

New member
Dec 25, 2013
6
0
0
Well, thanks for the detailed information!

If I understand your post correctly, you did not compare TOSLINK vs. USB from the same two devices, right? You actually introduced another device in the chain, the Raspberry Pi—and then your comparison was with and without this added device. Don't you think that that could be the source of the sound differences? The HiFiBerry project was born due to the defficiencies of the Raspberry Pi for Hi-Fi. Have you tried that?

A few weeks back and due to another thread in this forum, I did a quick comparison of lossless 44.1KHz/16-bit audio from the computer to the Airport Express through the network, and then to the Arcam irDac through TOSLINK, vs. directly from the computer to the irDac through USB, and could not find differences. USB allows for higher-resolution audio and avoids the buffer in the Airport Express—so the music starts quicker once you play it in the computer. But sound-wise and specifically with lossless 44.1KHz/16-bit audio, I did not find differences.

Have you maybe also performed this test without the Raspberry Pi?
 

BMFDrums

New member
Dec 15, 2012
8
0
0
You are exactly correct - I did not compare Toslink vs USB from the same device, as that would have required moving my computer, and that is not something that will be the final solution, so I didn't check that. I probably should have mentioned that the Airport Express cannot output audio via USB (the USB port is for connection to printers, etc. only), and the RPi does not have a Toslink connection as standard, so I couldn't do a like-for-like comparison with those devices.

I have tried the HiFiBerry - one of the links to the Volumio forum that I posted above is about my experience with the HiFiBerry, and also here on What HiFi: http://www.whathifi.com/forum/hi-fi/my-raspberry-pi-hifiberry-music-server#comment-2933830

I was extremely impressed with the sound that such a simple and relatively simple DAC can produce, although I am replacing the RPi with CuBox-i2 Ultra 'servers' to enable USB.

About your experiment with your AE and computer into the irDac, it is also not a 'perfect experiment' (no criticism intended - just an observation), as you have also changed multiple elements at once. What would be a fairer comparison of USB vs Toslink would be to connect your irDac directly to your computer through both Toslink and USB.

If I understand correctly, what you have done, though, is provide contradictory evidence to what I suggested - that you observe no difference in computer -> router -> AE -> Toslink-> irDac vs computer -> USB ->irDac. That's interesting in itself, so maybe the advice to others is "maybe it depends on the set-up, so you may have no choice but to experiment". That's not ideal, but may be just a reflection of specific combinations of source, network, DAC, etc. Out of interest, did you try a track with very low music frequencies? If not, I would urge you to try again with the track I suggested - James Blake's "Limit to Your Love". The notes of the sub-bass were clearly different (better) for me using USB (admittedly from the RPi, not the AE).

Just a final point on your comment "USB allows for higher-resolution audio". That is true in your test system, as the AE (and specifically Apple Airplay) is limited to 16/44, although connecting your computer directly to your irDac using Toslink (if possible) can also deliver hi-res files. If you were testing both with CD-quality 16/44, that is a fair comparison of AE vs computer. If you had used a hi-res file on the computer and streamed it to the irDac using the AE, it would have been down-sampled to 16/44 by Airplay, so that would not have been a fair comparion. But, you didn't do that - well done!

Thanks for your feedback - always appreciated.
 

unsleepable

New member
Dec 25, 2013
6
0
0
Well, I didn't intend to be scientific with my test. Honestly, I just wanted to see if I got better audio quality through USB than through TOSLINK; if the Airport Express was reducing the quality of the audio reaching the irDac. And my conclussion so far is that it doesn't.

I have just tried with the track that you suggested—although from Spotify Premium, since I don't have that particular disc. Again, all the same to me through USB and Airport Express/TOSLINK. I would suggest it could be jitter in your Raspberry Pi. By the way, interesting track.

If you are into hardware and programming, the following is a good read to understand the implementation of S/PDIF audio and have a better understanding of the whole issue with jitter. S/PDIF Digital Audio on a Microcontroller.
 

BMFDrums

New member
Dec 15, 2012
8
0
0
Hi, Ifor

No, the CuBox does not have a built-in DAC - I will only be using them as servers to my existing async-USB DACs (Audiolab 8200CDQ and M-DAC). They do have S/PDIF (Toslink) outputs, although my previous experience suggests that in my set-up USB sounds better than Toslink (admittedly using Apple Airport, not the CuBox). Note that this is not what unsleepable found - his experience was that he found no difference between USB and Toslink. I will experiment with Toslink as well as the USB connection, and use whichever sounds best to me.
 

ifor

Well-known member
Dec 3, 2002
114
12
18,595
BMFDrums said:
Hi, Ifor

No, the CuBox does not have a built-in DAC - I will only be using them as servers to my existing async-USB DACs (Audiolab 8200CDQ and M-DAC). They do have S/PDIF (Toslink) outputs, although my previous experience suggests that in my set-up USB sounds better than Toslink (admittedly using Apple Airport, not the CuBox). Note that this is not what unsleepable found - his experience was that he found no difference between USB and Toslink. I will experiment with Toslink as well as the USB connection, and use whichever sounds best to me.

i look forward to your feedback. I have been considering a Pi + HiFiBerry Digi rather than HiFiBerry DAC, but CuBox is an option.
 

BMFDrums

New member
Dec 15, 2012
8
0
0
ifor said:
i look forward to your feedback. I have been considering a Pi + HiFiBerry Digi rather than HiFiBerry DAC, but CuBox is an option.

It will be a while before that can happen - CuBox(es!) not due to arrive until late April :cry:
 

andyjm

New member
Jul 20, 2012
15
3
0
BMFDrums said:
Toslink provides electrical isolation of the DAC from the source, but the clock controlling the data is in the 'push' device, normally a relatively poor quality device (leading to higher levels of 'jitter' when compared with the clock in DAC); asynchronous USB transfers the control of data flow to the (normally much better) DAC clock (reducing levels of jitter), but the two devices are (normally) connected electronically, so any electronic interference from the source device can affect the DAC. I say 'normally' for the USB connection, as there are power supplies available that provide the DAC with a separate ultra-linear DC supply that is isolated from the supply to the computer/streamer.

Correct about TOSLINK not so about USB.

Even if the DAC has a separate power supply, most DACs will still be galvanically connected to the PC if using a USB cable. There are exceptions, there is a USB isolation device designed for use in medical equipment that provides full galvanic isolation - I will see if I can remember who uses it.
 

TRENDING THREADS

Latest posts