Question for Mr. Smith please. About toslink

Interesting. Despite its adherents, have never felt it sonically. It has been a standard for getting 8 channels of digital audio from Device A to B in pro gear for a while now. Never mind two channels. One of those places where pro and Audiophile audio perspective clashes.

TOSLink only does 26/96 by spec. But most devices can do better and most cables that arenā€™t free (e.g. cables that donā€™t come with a DVD player, etc.) will do 24/176.4 or 24/192.

With the DS itā€™s worth an experiment. Also it might be worth trying TOSLink with all other inputs unplugged to hear the effect of the galvanic isolation in your system, even if you donā€™t plan on using TOSLink that way.

Not my preferred way to get to galvanic isolation sonically. And it sorta creeps me out to say, ā€œThis gauge of firehose May Be Adequate to put your house fire outā€ ; )

Toslink, in my experience, sounds thin and brittle.

Badbeefā€¦ Transparentā€¦ me too, I now use a few of their speaker cables for two systems I have. I had a chance to audition many used cables at once across a wide range of pricesā€¦ The Transparent cables always sounded betterā€¦ but not more detailed, not more whateverā€¦ If I chose an aspect to evaluate, such as upper range detail, Transparent never ā€œwonā€. But overall, their cables just are good to me.

Peace
Bruce in Philly

Transparent is top of the line. I do not like that they do not give specs. They are obviously protecting secrets. If you call them they will give specs though. I would only buy it used. The reason I could afford to buy it new is I buy it used! My most expensive Transparent was $32,000 and I paid $1,250. It is worth nothing used. Plus it is already broken in! The coax is not a firehose. Their TOTL speaker cable is. equivalent of 12x4awg. That was my expensive one. 12 feet each side with 3 mystery boxes and elevators. I also feel Crystal cable is very good but it holds itā€™s value much better. For whatever reason I donā€™t know.

However for Power I prefer AC-12. Which is a no brainer going out of power plant 20. The DS is not TSS but it is very good. The power plant 20 is maybe akin to TSS. It is the best power delivery I ever heard. Power goes out often. 2 minutes until generator turns on with huge spike. Light bulbs have shattered but my audio is protected.

I would certainly not overlook cables and power. Even if you only have a $20,000 system. At well into six figures it is a must here. It might seem expensive but maybe many people do not need the 20. Certainly never ever ā€œgangā€ power cords! Get as many outlets as you need.

Transparent used is the biggest deal in audio hobby IMO.

You have a sense for finding cable bargains. I donā€™t usually buy used stuff, but if you can find a hi-end used 5m coax with RCA on either end please let me know. Or a multi-strand Toslink will do. But must be a good one.

Try audioconsultants.com used page. They may still have some lovely cables for nothing. Great store. They love to sell you for $$$ and give you $ on exchange LOL. When I looked at the Magico he offered but $20,000 for my speakers.

First of all the Transparent digital link XL 1M is not $10,000 as I stated.
It is a mere $3,195.

That being said. They state it uses ā€œnon metallicā€ materials. However that is possible. Therefore I wonder if it might have the galvanic isolation of Toslink? I guess one would only know if they knew exactly what the cable is comprised of? I cannot see how non metallic cables transfer electricity though?

I can get a great deal on a very high end Toslink so as MR Smith stated I might as well get that too. In fact this Toslink is so good it states it can transfer up to 5.2MHZ! If that is to be believed either. It should do 384/24 and DSD then. It does not even matter because my CD transport only outputs at 44.1/16.

A really good Toslink cable is only as good as the Toslink ports you plug it into, which are usually $3 parts ; ). Youā€™re pretty much always converting from and to i2s, whether itā€™s Toslink, coax, AES, USB - hence Tedā€™s approach.

1 Like

What exactly is MR. Smithā€™s approach? I thought he just prefers a Toslink from CD? I am not quoting you MR. Smith because I am not sure.

It seems most other people prefer Coax. I have a $3,195 Coax cable VS $275 Toslink I can get for $20 used. It is 1380 strands of plastic. This guy prefers plastic over glass. In fact ATT uses plastic for miles to carry thousands of GB per cable but their cables are 10ā€ in diameter. I personally feel Tos is not as good for why you say. I have to drive 160 miles to get it. Worth drive?

I was also wondering what you guyā€™s think of IFI IGalvanic while I am up there? I use MR. Smiths approach. High end USB hub with +/- cut on cable and Linear PSU to hub. IGalvanic worth it? It uses Hub chip in fact. I know now we are talking USB but I figured if I make the driveā€¦.

I was referring to his approach of usually (not in the case of the two-box TSS, I guess) keeping it i2s as much as possible, including accepting i2s input on the DS DACs, etc., to avoid conversion to anything else. After the DAC, yeah - you make yer choicesā€¦

Some people seem to prefer the sound of Toslink in their systems. I have never cared for it myself, either glass or plastic. Itā€™s one of those things where, just because it was designed to carry lots of phone calls lots of miles in no way means it is the best way to move digital audio 1 meter between sensitive audio devices. Completely different thing. Unless you are a ā€œbits are bitsā€ adherent, in which case you have nothing to worry about and it doesnā€™t matter how much you spend.

AHH, I will keep my 20 bucks and $150 of gas LOL.

What about Igalvanic? Worth it with what I already have going? I have no power lines but noise on data lines?

Not familiar with it.

Thatā€™s not quite right in terms of how the DS works.

In its original form, I2S uses three wires: a bit clock, a L/R word clock and a serial data line. In contrast, the SPDIF variants including Toslink use a binary encoding scheme on a single conductor/fibre. SPDIF sends ā€œframesā€ of data which have 24 bits space for a single PCM sample and 4 bits of other info including which channel the sample is for. The clock is not explicit ā€“ however fast the data arrives, thatā€™s your clock. You have to derive your audio clock by syncing up with the edges of the SPDIF frame transitions.

Youā€™re right that most DACs decode SPDIF to I2S. The DAC chip itself often has just a single input that takes I2S format. Thatā€™s not how the DS works.

There are four two-wire I2S inputs to the FPGA and three single-wire SPDIF inputs. All of them are permanently connected and able to be observed by Tedā€™s code. We just pay attention to only one input at any given moment.

The I2S inputs come from the two HDMI connectors, the USB receiver chip and the Bridge slot. They are two-wire only: L/R and serial data. The bit clock is not needed because of how the DS works, which Iā€™ll talk about in a minute. For PCM or DoP the audio data is all on the serial data line and the L/R clock marks the start of a new sample for left and right channels alternatively. For native DSD each wire carries one channelā€™s bitstream.

The SPDIF inputs are of course from the coax, XLR and optical connectors. The optical receiver converts the light pulses into electrical ones but otherwise does not transform or decode the data in any way.

The FPGA runs at a clock speed vastly higher than any of the input signals. Because of this it is able to analyse the incoming signals in an abstract and mathematical way to extract their audio data for processing, instead of being almost mechanically coupled to the incoming stream the way traditional PLLs and DAC chips are.

On an I2S input it recognises the difference between PCM/DoP and native DSD data based on the oscillations of the L/R clock line. Multi-bit samples will be pulled alternatively from the serial data line into the L/R processing pipelines. Native DSD data just hops bit by bit from the wire into the appropriate channelā€™s pipeline.

All the SPDIF inputs are single wires. The FPGA can directly detect the ā€œstart of frameā€ preamble, from that derive the transmission rate, and then extract the frame with its header and its 24 bits of audio data. All of that is done in software without any conversion to I2S.

It doesnā€™t matter what sample rate the incoming data runs at, the FPGA always runs at the same speed and places samples into its internal processing stream in a regular fashion for upsampling to 20xDSD rate and 50+ bits of resolution.

This same mechanism is also key to the DSā€™s impressive immunity to source jitter. The FPGAā€™s input sampling buffer has a bit of wiggle room, so the inputs can arrive with somewhat inconsistent timing and it makes no difference to the timing that the FPGA uses for its processing pipeline. If the source is on average a bit slower or a bit faster than the FPGA is operating, the FPGA just gently adjusts the overall speed of the DSā€™s master clock to match.

1 Like

Ted, is there something specific other than ā€œspecā€ holding back the DS Sr from accepting more than 24/96? I know the optical output on my Asus X99 Deluxe motherboard will do 24/192. Iā€™ve done it before with a NAD M51 DAC and modestly priced glass optical cable I own without issues. I also noticed the DS Sr will accept a 24/96 signal but not 24/88.1 over optical. Is there a reason for this?

Most non-cheapo cables and most TOSLink transmitters and receivers support 176.4 and 192k fine. 24/88.1 works just fine on the DS. Many motherboard drivers have quirks - if you are using ASIO try WASAPI.

Have no idea if what you say isā€¦whatever.

Will leave it to Ted, if he has the time.

Sorry, I do tend to dive deep. All that info about the DS is what Iā€™ve learned from Tedā€™s posts, videos and a couple of emails. Hopefully I didnā€™t get anything very wrong.

Point being: the Toslink input gets converted from light pulses to electrical voltages but never to I2S. The FPGA knows how to directly extract the audio samples from the SPDIF signal without an I2S conversion.