Toslink with the DirectStream DAC


#1

Hi again Ted @tedsmith

I know you prefer the Toslink input for the many excellent reasons like:

  1. immune to RFI & EMI

  2. galvanic isolation

  3. jitter a non-issue due to elimination by the DSD

  4. straight wire path from the Toslink input to the FPGA, etc. I hope that covers all the pro’s of Toslink :slight_smile:

I’ve been reading about losses in plastic vs glass. Like all other cables in this hobby, things can get expensive very quickly.

But what do you look for in a good Toslink cable?

More importantly, which Toslink cable do you prefer to use?

Many thanks again :slight_smile:


#2

To get 24/192 you may need a glass cable, I have a plastic cable that allows the DS to receiver 24/176.4 reliably but fails to pass 24/192k. Glass doesn’t have to be terribly expensive. I have a bottom of the line AudioQuest glass cable from many years ago (they change the names so often that even if I remembered the name it wouldn’t do you a lot of good.) I suspect that any TOSLink cable that’s made well with well polished ends will work fine at 24/192 and that any TOSLink cable that came with a satellite receiver, cable box, etc. won’t :slight_smile:


#3
Ted Smith said

To get 24/192 you may need a glass cable, I have a plastic cable that allows the DS to receiver 24/176.4 reliably but fails to pass 24/192k. Glass doesn’t have to be terribly expensive. I have a bottom of the line AudioQuest glass cable from many years ago (they change the names so often that even if I remembered the name it wouldn’t do you a lot of good. I suspect that any TOSLink cable that’s made well with well polished ends will work fine at 24/192 and that any TOSLink cable that came with a satellite receiver, cable box, etc. won’t :slight_smile:


Thanks Ted!

#4

I have read good things about the Lifatec glass cables. I intend buying the DirectStream DAC and will get a Lifatec cable for it.


#5

The optical receiver in the DS is only guaranteed to 96kHz sample rate for stereo SPDIF signalling… but physical quality of that part varies in manufacture and some parts exceed the spec so that they’ll work at higher rates given a clear-enough signal.

My DS reliably works at 192kHz using even very cheap plastic Toslink cables up to about 3 feet long. It sounds fabulous. I can’t think of any reason why glass would sound different to plastic so I’ve not tried anything more expensive.


#6

I experimented with Toslink cables some years ago (for less than hi-res signals, using it to feed my DVR into my PWD MK 2) and found that I preferred the sound of glass, it had a more open and dynamic sound. I then compared two glass cables and Lifatec won for me, I have it in use for my DVR feed into my DS and the sound is excellent. (24 bit in).


#7

Mi2016 - Kind of a thing you need to try for yourself. I personally haven’t heard an optical link that I preferred over other means (when available). The Lifatecs do sound good/better IMO - one of the things with some plastic cables (as Ted alludes to) is the ends aren’t always cut cleanly/flat and not always polished.


#8

I have had good luck with Toslink cables that use medical grade fiber. I have a 2m IXOS (hard to find) and 1m Analysis Plus that can both pass 24/192 to my DirectStream DAC, whereas my other Toslink cables can’t pass that bit rate or stutter at that rate. My theory is that a Toslink cable that can pass 24/192 is good quality and neither of these is real expensive. Plus they are not as fragile as glass.


#9

I don’t dispute that people have heard differing results from different optical cables on other DACs. The thing is - why would that make a difference, and does that effect apply to the DirectStream DAC?

The answer to the second question, as far as I can ascertain, is no!

An optical cable simply guides light from the transmitter to the receiver. The receiver is watching for transitions between bright and dim states, which it then mimics with voltage transitions between high and low on its electrical back end. The cable needs to be clear enough that a bright state can be unambiguously detected, and it needs to minimise internal reflections so that when the transmitter stops shining the cable becomes dark as quickly as possible.

The SPDIF signalling protocol is cheap and easy to implement but has a huge flaw when it comes to clocking. In most cases the sender uses its own clock and the receiver is forced to sync itself to the arrival timing of the incoming high/low or bright/dim transitions. Just about every DAC prior to the DirectStream uses phase-locked loops or asynchronous sample rate conversion to process the incoming signal stream. PLLs try to make the DAC follow the timing of the signal as if it’s a conductor, while ASRC uses mathematical approximation to map the incoming stream with reference to a local clock at a higher resolution. Both of these approaches are impacted in audible ways by variations in the time between each successive signal pulse seen from the electrical output of the optical receiver. That is to say, input jitter matters in these DACs.

To reduce input jitter, you need a good stable clock at the source end to start with. That’s one main reason why different transports or other digital sources sound different with SPDIF connections. Then with optical you need the receiver to be noticing and reacting to the bright/dim transitions with precise regularity. If the light doesn’t shine brightly enough it’ll be slow to respond to a bright state, and if there’s too much scattered reflection it’ll be slow to respond to a dim state. That’s why the quality of your optical cable matters with those DACs.

The DirectStream is different to anything which came before it. It has a single ultra high stability clock of its own at a frequency which is a high integer multiple of all the supported digital audio sample rates - it’s up in the MHz somewhere, I forget the exact value. The stability is not about long-term average but about consistency from one clock pulse to the next, and in fact the overall rate is adjustable by varying the voltage applied to the clock. That clock directly drives the output stage but it also controls the FPGA and all the input processing.

The FPGA input processing watches the incoming signal line for transitions between high/low states. That doesn’t mean it reacts to such changes, rather it’s making millions of observations every second to see whether the signal is currently high or low. Based on the pattern of changes it sees, and the spacing between them, it can identify SPDIF bi-phase mark encoding and start to extract the data. But here’s the most beautiful thing: it does this all so fast that it can notice if it’s drifting away from sync with the incoming data stream, and it gently adjusts the voltage going to the ultra stable master clock so that it matches the overall pace of the source.

What this means is that the kinds of jitter which make Toslink generally the worst option for other DACs and which make Toslink cable quality an issue for those DACs are basically irrelevant for the DirectStream. A really unstable source will see the DS having to make repeated adjustments, impacting the sound. But given reasonable quality sources and cables which aren’t right on the edge of the functional envelope, optical is a fantastic and cheap route to great quality sound from your DS.


#10

Hi guys, great info shared all round here.

But my original question was about the losses that are sometimes mentioned between glass and plastic Toslink cables, with glass being less lossy.

If a plastic Toslink cable plays 24/192 music un-interrupted, is it safe to say there are zero losses?

Or are there still losses that are happening that may affect the sound quality, even if the 24/192 music plays un-interrupted with the plastic cable?

Maybe Ted can help, since this is technical @tedsmith

And yes, I know it’s all about ‘trusting your ears’ but like many here, I have an interest in the technical stuff, even if some of it is way over my head lol

Cheers again guys


#11

Transmitters and receivers are built assuming a certain amount of signal loss, but remember this a lower level signal (within reason) doesn’t change the bit error rate noticeably. The advantage of less lossy cables is that they can be longer before becoming unreliable.

Errors sending and receiving bits in digital audio interconnects (or USB or Ethernet for that matter) are virtually non-existent in a nominally working audio system. If you want to know if your system is getting any errors because of some problem use the bit-perfect test: it will fail with ANY bit error during the test - http://www.psaudio.com/ps_how/how-to-run-a-bit-perfect-test-with-directstream/ (If your system plays DSD via DoP over the TOSLink then it likely is receiving 24/176.4 reliably, any big errors in the top 8 bits of DoP will cause a noticeable dropout where the DAC assumes that the signal has reverted to PCM.)


#12
Ted Smith said

Transmitters and receivers are built assuming a certain amount of signal loss, but remember this a lower level signal (within reason) doesn’t change the bit error rate noticeably. The advantage of less lossy cables is that they can be longer before becoming unreliable.

Errors sending and receiving bits in digital audio interconnects (or USB or Ethernet for that matter) are virtually non-existent in a nominally working audio system. If you want to know if your system is getting any errors because of some problem use the bit-perfect test: it will fail with ANY bit error during the test - http://www.psaudio.com/ps_how/how-to-run-a-bit-perfect-test-with-directstream/ (If your system plays DSD via DoP over the TOSLink then it likely is receiving 24/176.4 reliably, any big errors in the top 8 bits of DoP will cause a noticeable dropout where the DAC assumes that the signal has reverted to PCM.)


Fantastic, thanks for the technical explanation Ted. I’ll be sure to run the test when I get home.

#13

I don’t want to be ignorant, but what could be source components for a Toslink connection with a sampling rate greater than 44.1 or 48?

My higher sampling rates are from my computer either using USB or the bridge into my DSJr.


#14

Hi mate, just one example only but my Apple iMac outputs up to 24/192 via it’s mini Toslink output.


#15
Deanhorn said

I don’t want to be ignorant, but what could be source components for a Toslink connection with a sampling rate greater than 44.1 or 48?

My higher sampling rates are from my computer either using USB or the bridge into my DSJr.


Many DVD, DVD-A, BluRay, etc. transports will put out 24/88.2 or 24/96 if the source 24/88.2 or 24/96 and it isn’t DRM’ed or otherwise copy protected (DVD’s and DVD-A, etc. can choose to limit their output to, say, 48k even if they are 24/96…) 24/192 is rarer but out there.

The USB -> TOSLink converter(s) I have will try to do 24/192 if that’s what the source is.

The TOSLink connections on my computer and docking station, etc. do 24/96…


#16

SqueezeBox Touch with EDO extension installed does 192 over USB, coax and optical – and Roon supports SqueezeBox as an output. It’s a pretty nice fuss-free setup.


#17

The Sonicorber SE and HiFiBerry Digi+ also output up to 24/192 over Toslink.


#18

The Auralic streamers also output 24/192 via Toslink.

Back on the topic of which cable, has anyone tried Wireworld Supernova glass Toslink? I think 7 is the current iteration.

Any feedback on it compared to the Lifatec?

Thanks


#19

Hey folks. Is this a good thread for crow-eating?

Nice and quick: for a long time the optical input was the way I got the best sound out of my DS, and I was very happy to tell everybody about it being awesome. So I feel obliged to post that I’ve been won over by the Curious USB cable as the superior connection in my system.

As good as the DS’s jitter rejection is, it seems that having the DAC in control of the clock overall is still a winner. And whatever the Curious cable does to block the electrical grunge which had brought the USB input undone using other cables… works well enough to counteract the total isolation of Toslink.

The quest for perfect sound continues. itwasntme_gif


#20

The Curious USB is really really good.

Having the power line separated and shielded from the data/ground lines is one big advantage - they’re not the only ones doing this.

I think their shielding of the data/ground is different to others. And they said they’ve paid close attention to grounding but haven’t gone into details. They have to keep some things secret don’t they !?

All of this in combination with the pure silver conductors for the data lines (better signal integrity?) and maybe other design differences to others.

I’ve tried a lot of different USB cables. I used to be a ‘bits are just bits’ guy but I think all of the above points help to reduce the effects of interference with the USB signals, which we know do affect sound quality.

Whatever they are doing, the end result is something special though.