SPDIF Optical Cables, can they perform different?

[simplywyn]

I have a cheapo SPDIF audio cable from amazon, and thick, braided fancy metal / gold ended SPDIF cable that someone gave me when I bought the Topping D90 off them.

When I plug in my Amazon SPDIF cable and the other Gold ended one, the gold ended one seems to sound better.

As an engineer, I think this is complete horseshit as I've work with optical cables and it has nothing to do with audio quality, basically you either get it or you don't.

Anyone know something I don't here? As far as I'm concerned, this is the HDMI problem, a $5000 HDMI cable doesn't perform any better than a $5.

Anyone care to tell me I'm an idiot for thinking it sounds better (and prove that SPDIF cables matter?)

 

[antcollinet]

It doesn't sound different.

Unless the cheap one is so bad data is being lost cauing pops/crackles, then the sound is identical.

The only exception might be if the receiving device is so poorly designed that it uses the recovered clock from the data stream to clock the DAC - thus causing excessive jitter, and then only if the optical fibre in the cheap cable is so crap that it significantly distorts the optical signal. Certainly braid/gold/metal can made no difference for an optical cable.

Any decent DAC (even the cheap ones) generate a local stable clock for clocking the DAC, thus effectively eliminating jitter.

 

[DVDdoug]

When I plug in my Amazon SPDIF cable and the other Gold ended one, the gold ended one seems to sound better.

"Placebo effect". What is a blind ABX test?

As an engineer, I think this is complete horseshit as I've work with optical cables and it has nothing to do with audio quality, basically you either get it or you don't.

Right. It's digital. The data can get corrupted but if that happens it usually obvious and you wouldn't just say "sounds better", you'd say "I'm hearing crackles", or something specific. When you get a digital error, a big error is just as likely as a small error. If a bit gets flipped in your bank's computer you are just as likely to see a 1-cent error as a million dollar error.

And with digital, it's impossible to accidently loose bass, or loose volume, or anything "analog" like that.

 

[radix]

Did you mean the TOSLINK optical cable or the SPDIF RCA cable? It's the same data, just different cable. Really, it's all pretty much the same between AES, spdif, and toslink over XLR, BNC, and RCA. There's some differences, but generally AES equipment understand spdif just fine.

Anyway, like @tonycolinet says, I think as long as the cable is in spec, it will be just fine.

Marc

 

[shal]

Hi,

Have you a statement about some sentence made by Dan Lavry ( http://www.lavryengineering.com ) ?
https://gearspace.com/board/high-end/335502-spdif-aes-compatibility-lavry-da.html


Quote:

There is a reason for that: impedance matching to avoid signal reflections on the cables. When the cable is short, the signal propagation time is short (relative to the rise and fall times of the digital signal), and the reflection issue becomes insignificant. When the cable propagation is long (relative to rise and fall times), then impedance matching is in order.
SPDIF signals are weak, so if possible, use a real short cable (such as 3 feet is better then 6 feet). The shorter the better.

 

[pozz]

I have a cheapo SPDIF audio cable from amazon, and thick, braided fancy metal / gold ended SPDIF cable that someone gave me when I bought the Topping D90 off them.

When I plug in my Amazon SPDIF cable and the other Gold ended one, the gold ended one seems to sound better.

As an engineer, I think this is complete horseshit as I've work with optical cables and it has nothing to do with audio quality, basically you either get it or you don't.

Anyone know something I don't here? As far as I'm concerned, this is the HDMI problem, a $5000 HDMI cable doesn't perform any better than a $5.

Anyone care to tell me I'm an idiot for thinking it sounds better (and prove that SPDIF cables matter?)

Build quality and consistency, reliability of connectors, shielding, rating for distance matter. All specs you can get from professional companies. I buy from them for that reason.

On this front page you can find a list of articles about digital audio, cables, etc.: https://www.audiosciencereview.com/forum/index.php?threads/master-review-index.11398/ Bottom line is there's nothing to hear unless it's an obvious problem, like a ground loop, which with optical is not an issue given that it doesn't electrically connect devices. Only issue would be drops in audio, stuttering.

 

[Berwhale]

Optical cables can either be glass fibre or plastic (POF), my understanding that either should be fine for short distances (i.e. in a domestic setting), but glass may be required for longer runs.

 

[solderdude]

Anyone care to tell me I'm an idiot for thinking it sounds better

There are definitely measurable differences in TOSLINK cables and connections
That doesn't mean they will sound different as there is no analog audio going through those cables.
To test the latter have someone else randomly plug in one of the cables while you do not know which cable is connected.
Then use your ears and see if you can still reliably tell audible differences.

 

[Blumlein 88]

There can be differences. For instance if you get into rather long Toslink cables you may find some work at higher sample rates and others will sometimes glitch and have a dropout. OTOH, if you don't get such glitching there shouldn't be a difference in sound quality. I think the official view is Toslink should be 5 meters or less. I've used cables twice that long which work fine however.

 

[VintageFlanker]

There are definitely measurable differences in TOSLINK cables and connections

You mean: between two different Toslink cables?

I would like to see published measurements about that.

For now, @Archimago found the opposite: http://archimago.blogspot.com/2013/05/measurements-toslink-optical-audio.html?m=1

 

[LTig]

There can be differences. For instance if you get into rather long Toslink cables you may find some work at higher sample rates and others will sometimes glitch and have a dropout. OTOH, if you don't get such glitching there shouldn't be a difference in sound quality. I think the official view is Toslink should be 5 meters or less. I've used cables twice that long which work fine however.

5 m should work always. Longer connections can work depending on the type of transmitter and receiver.

 

[RayDunzl]

Optical cables can either be glass fibre or plastic (POF), my understanding that either should be fine for short distances (i.e. in a domestic setting), but glass may be required for longer runs.

Longer, with a powerful transmit laser, and Corning glass fiber, was up to 60 miles when I last fooled with it in 2002.

The signal would becone distorted at that distance, but run it through a short length of fiber with the opposite dispersion characteristics, and it would be fixed up well enough to apply some form of Rahman Amplification and shoot it down the road (often rail tracks) to the next amplifier, or to the receiver which splits out the transmitted wavelengths (channels) and converts to an electrical signal for switching, or could be retransmitted as a fresh undistorted pulse train on a long run, like LA to Chicago.


I'm sure you're talking terrabit rates now. We were doing 32 channels and a half-terrabit.

Yup.

Seven years ago:
255Tbps: World’s fastest network could carry all of the internet’s traffic on a single fiber

255Tbps: World's fastest network could carry all of the internet's traffic on a single fiber

A joint group of researchers from the Netherlands and the US have smashed the world speed record for a fiber network, pushing 255 terabits per second down a single strand of glass fiber. This is equivalent to around 32 terabytes per second -- enough to transfer a 1GB movie in two milliseconds...

www.extremetech.com

Glass is "better" than plastic, but no need at home.

 

[graphical]

Appreciate the perspective, RayDunzl.

Longer, with a powerful transmit laser, and Corning glass fiber, was up to 60 miles when I last fooled with it in 2002.

The signal would becone distorted at that distance, but run it through a short length of fiber with the opposite dispersion characteristics, and it would be fixed up well enough to apply some form of Rahman Amplification and shoot it down the road (often rail tracks) to the next amplifier, or to the receiver which splits out the transmitted wavelengths (channels) and converts to an electrical signal for switching, or could be retransmitted as a fresh undistorted pulse train on a long run, like LA to Chicago.


I'm sure you're talking terrabit rates now. We were doing 32 channels and a half-terrabit.

Yup.

Seven years ago:
255Tbps: World’s fastest network could carry all of the internet’s traffic on a single fiber

255Tbps: World's fastest network could carry all of the internet's traffic on a single fiber

A joint group of researchers from the Netherlands and the US have smashed the world speed record for a fiber network, pushing 255 terabits per second down a single strand of glass fiber. This is equivalent to around 32 terabytes per second -- enough to transfer a 1GB movie in two milliseconds...

www.extremetech.com

Glass is "better" than plastic, but no need at home.

Appreciate the perspective, RayDnuzl. I was reading this thinking about fiber loops from my Bell Labs days. Quality of the glass made a difference. The distortion would begin to harm the SNR once you got out there a couple of miles. Not really a concern at home.

But, I was always curious. I don't know the SPDIF protocol, but why didn't they include error correction? Would it have made the transceiver too expensive?

 

[Blumlein 88]

Longer, with a powerful transmit laser, and Corning glass fiber, was up to 60 miles when I last fooled with it in 2002.

The signal would becone distorted at that distance, but run it through a short length of fiber with the opposite dispersion characteristics, and it would be fixed up well enough to apply some form of Rahman Amplification and shoot it down the road (often rail tracks) to the next amplifier, or to the receiver which splits out the transmitted wavelengths (channels) and converts to an electrical signal for switching, or could be retransmitted as a fresh undistorted pulse train on a long run, like LA to Chicago.


I'm sure you're talking terrabit rates now. We were doing 32 channels and a half-terrabit.

Yup.

Seven years ago:
255Tbps: World’s fastest network could carry all of the internet’s traffic on a single fiber

255Tbps: World's fastest network could carry all of the internet's traffic on a single fiber

A joint group of researchers from the Netherlands and the US have smashed the world speed record for a fiber network, pushing 255 terabits per second down a single strand of glass fiber. This is equivalent to around 32 terabytes per second -- enough to transfer a 1GB movie in two milliseconds...

www.extremetech.com

Glass is "better" than plastic, but no need at home.

In just a few years this will be the minimum acceptable connection for DSD 16,384 playback.

 

[pkane]

In just a few years this will be the minimum acceptable connection for DSD 16,384 playback.

That'll work very well with my 1000 petabyte music NAS!

 

[RayDunzl]

I don't know the SPDIF protocol, but why didn't they include error correction?

It is a real time unidirectional stream.

The DAC adjusts its clock rate to match the rate of the incoming data.

https://www.st.com/resource/en/application_note/an5073-receiving-spdif-audio-stream-with-the-stm32f4f7h7-series-stmicroelectronics.pdf

What errors do you hear?

 

[radix]

But, I was always curious. I don't know the SPDIF protocol, but why didn't they include error correction? Would it have made the transceiver too expensive?

AES has a single parity bit per 16-24 bits of audio per channel. AES uses very small sub-frames of 32 bits per sample per channel, so 64 bits for a full frame of L+R content. Professional AES supports up to 50 kHz frame rate at 100m over balanced twisted pair. The standard says higher frame rates or longer distances are possible with proper cable selection and equalization. 50 kHz frame rate = 3.2 Mbps, so it requires a 6.4 Mbps clock. 48 kHz 24-bit audio would need a 6.1 MHz clock (I rounded a little). If you want 192 kHz 24-bit audio, you need a 24.6 MHz clock (I rounded a little).

It's been a long time since I did frequency response calculates for square-wave signaling over copper, but if I remember correctly, over 20 - 30 MHz starts getting a fair bit of attenuation. You could look at an attenuation chart for overall loss, but I think the main issue is the rise/fall timing.

Anyway, I suspect that they engineered the SNR and rise/fall times such that the BER is so small that a larger CRC was not called for, at least in the normally envisioned signalling rates. Using FEC would add significant delay and a lot of bit overhead. AES is a simplex link, so there is no error correction possible (apart from FEC).

Marc

 

[simplywyn]

5 m should work always. Longer connections can work depending on the type of transmitter and receiver.

Yes my work in signal attenuation over fiber optic tells me what I know about distance. But again, the results are either it WORKS or it DOES NOT WORK and not, it sounds better or worse.

You don't get slightly better signal, you either get signal that works, or signal that is just full of noise and totally unusable (basically the square waves become noisy to a point where you can't determine the 1-0 changes)

 

[Doodski]

(basically the square waves become noisy to a point where you can't determine the 1-0 changes)

Indecisive waveforms are what I call them. That's what I was taught to call them anyway.

 

[solderdude]

You mean: between two different Toslink cables?

I would like to see published measurements about that.

For now, @Archimago found the opposite: http://archimago.blogspot.com/2013/05/measurements-toslink-optical-audio.html?m=1

My daily work is in opto electronics and work with fiber optics all day.
Not 2 cables measure the same.
Unless attenuation becomes too much and data is corrupted the transmission of data is the same though.
Just as every different electrical cable will all measure differently (far outside audible range) yet may have no (audible) effect on the transmitted waveform.