Plastic vs glass fiber TOSLINK?

[drummerdimitri]

So I have a DAC connected to my computer via an optical plastic fiber TOSLINK cable that is 3m long.

After some research, I found that there are also glass optical fiber cables that attenuate the signal 10 times less than the plastic ones (1 dB/m) meaning that I would of lost 50% of the light energy by the time it reaches my DAC which seems like it would degrade the sound quality by a lot.

Would it make sense to switch over to a glass fiber version with the same length? I'm not too sure about the correlation of beam attenuation vs data loss at the other end.

[awallin]

one way of attenuating an optical signal in a fiber is to wrap the fiber around a pen or similar to get a tight bend-radius (which looses some of the light out of the fiber)
(see e.g. https://en.wikipedia.org/wiki/Optical_attenuator)

so the steps for an experiment would be:
- wrap fibre around a pen until you loose signal at the amp/speaker end
- back off just a bit so there is signal again
- listen if you can hear any difference

[extide]

As long as it's getting enough signal to decode the data the sound quality should be identical (literally bit-perfect). The only real thing you need to worry about in scenarios like this is jitter, which should not be affected by the cable at all.

[Benta]

As long as it's getting enough signal to decode the data the sound quality should be identical (literally bit-perfect). The only real thing you need to worry about in scenarios like this is jitter, which should not be affected by the cable at all.

This.

The TOSLINK interface is digital, so either you have connection or you don't.

[drummerdimitri]

As long as it's getting enough signal to decode the data the sound quality should be identical (literally bit-perfect). The only real thing you need to worry about in scenarios like this is jitter, which should not be affected by the cable at all.

This.

The TOSLINK interface is digital, so either you have connection or you don't.

I was looking to get this cable: https://www.qed.co.uk/cables/hdmi-digital/qed-reference-optical-quartz-digital-audio-cable.html#tab1

They don't seem to be selling BS and their claims of glass fibers reducing jitter is actually believable. Not sure if such a difference would translate to any audible benefits though.

[Benta]

No difference. The signal is re-synchronized at the receiving end, jitter wouldn't make a difference. And why would glass be different from POF on jitter?
Just a nice way of lightening your wallet.

[drummerdimitri]

No difference. The signal is re-synchronized at the receiving end, jitter wouldn't make a difference. And why would glass be different from POF on jitter?
Just a nice way of lightening your wallet.

I am convinced thanks for saving me 160 $ 

[helius]

The physical cause of jitter is optical dispersion, which affects both plastic and glass fibers. Dispersion is especially a problem for long lines (> 1 km), and so single-mode fibers are used to limit it. To my knowledge, only glass fibers can be made to be single-mode. At the low lengths and data rates used for audio it is not a particularly severe problem.

[jaycee]

Glass vs plastic fibre improving sound quality is complete BS. Simple as that.

If your DAC has such sensitivity to jitter that a change of cable influences the sound, it's a shit DAC. No good DAC should be using the received clock from SPDIF as the master clock, only using it as a reference to produce a clean clock. DAC's that use WM8804/5 as the SPDIF receiver for example.

[TheUnnamedNewbie]

The physical cause of jitter is optical dispersion, which affects both plastic and glass fibers. Dispersion is especially a problem for long lines (> 1 km), and so single-mode fibers are used to limit it. To my knowledge, only glass fibers can be made to be single-mode. At the low lengths and data rates used for audio it is not a particularly severe problem.

There is no physical reason that says only glass can be single-mode. The thing is that there is no real point making single-mode POF because you will probably still be limited by loss long before you are limited by dispersion. I actually thing single-mode fibers are likely cheaper to make than multi-mode, because of all the witchcraft they have to do with tapered refractive index cores and stuff to make the multi-mode fiber work decent. You just get that back in the fact that your connector is much, much cheaper because alignment is not so challenging.


Single-mode also only counters part of dispersion: multi-mode dispersion which you get in multi-mode fibers.  I doubt that SPDIF cables will be using single-mode fibers since the alignment requirements are just stupid expensive to actually reach. Single-mode fibers still suffer from dispersion due to two other reasons: the material themselves are dispersive and second is the nature of how the field distribution changes with wavelength in the fiber (I can go on about this but you get the picture). This is why dispersion-compensating fibers are a thing people need to achieve high datarate over long distances.

You can do some trickery and tune the profile such that the material dispersion and inherent field dispersion exactly cancel out at one (or two) frequencies, but that still results in only a narrow band you can work with.

[jaycee]

TOSLINK is designed to work with cheap plastic fibre, so single vs multimode is irrelevant. It's not even a laser - it's just a red LED being used!

[Berni]

No difference. The signal is re-synchronized at the receiving end, jitter wouldn't make a difference. And why would glass be different from POF on jitter?
Just a nice way of lightening your wallet.

Well that's not entirely true.

The audio master clock is actually recovered from the SPDIF signal on the receiving end. This clock is used to decode the encoded data stream and to drive the DAC. So a bad noisy input signal could cause extra jitter on the master clock that can affect the DAC performance.

That being said the reciever chip would only really have problems recovering a clean clock if the input signal is bad. Only form of signal deterioration on a fiber at these speeds is attenuation. So a glass fiber will only make a difference if you have a really long run of cable (Like 10 meters or more). Also when a receiver is right on the edge of having a usable signal it will likely start to cause bit errors, this can be clearly audible in the form of pops (Much like a dusty record). So if you can tightly wind the plastic cable (increasing attenuation) and not hear the pops then the receiver likely has enough signal strength to work with. For runs shorter than 5m this shouldn't be the case unless you have a crappy/damaged plastic cable.

[mikerj]

As long as it's getting enough signal to decode the data the sound quality should be identical (literally bit-perfect). The only real thing you need to worry about in scenarios like this is jitter, which should not be affected by the cable at all.

This.

The TOSLINK interface is digital, so either you have connection or you don't.

Absolutely not true, this is the kind of gross oversimplification I would expect to see from the average layperson rather than on eevblog.  SPDIF has no error correction (just a parity bit), if the link OSNR is degraded enough then multiple bit errors will occur and from experience they are very audible on an TOSLINK connection.

[extide]

Well, then it doesnt have enough signal to decode the data properly, does it

[TheUnnamedNewbie]

Well, then it doesnt have enough signal to decode the data properly, does it

The thing is that you seem to make it sound so binary - either there is enough signal to decode the data, or there isn't. In reality, it is a more gradual process. As you get a lower SNR, you start getting a higher bit-error rate (BER). A BER of 10^-12 might be common for good links, but it slowly tapers down as you get more and more errors. The 'either it works or it does not' is true for things like WiFi (and ethernet, to some extent, which has error detection and re-transmit) where error correction will make you pretty much never receive wrong data - if the link works, what comes out is correct. In the case of SPDIF, you can have a high bit error rate while still getting something out of your receiver, and as a result take a hit in sound.

[extide]

Ok, fair enough, but with a typical 3m toslink cable you shouldn't need to worry about any of that unless the cable is already damaged somehow.

[Bassman59]

So I have a DAC connected to my computer via an optical plastic fiber TOSLINK cable that is 3m long.

After some research, I found that there are also glass optical fiber cables that attenuate the signal 10 times less than the plastic ones (1 dB/m) meaning that I would of lost 50% of the light energy by the time it reaches my DAC which seems like it would degrade the sound quality by a lot.

Would it make sense to switch over to a glass fiber version with the same length? I'm not too sure about the correlation of beam attenuation vs data loss at the other end.

Given that my current design uses a gigabit-rate fiber link that the customer uses with a plastic fiber that's about 200 meters long, and it works perfectly, I think that a plastic TOSLINK fiber that's 3 meters long will work without problem.

[tooki]

I’ve used regular plastic 3m TOSLINK before, it works fine. (In fact, my understanding is that TOSLINK actually has a minimum length requirement!) I don’t doubt that glass fiber could go longer distances than plastic, but I suspect we’re talking about lengths that are orders of magnitude longer.

Oh yeah, you can also confidently ignore the cables with gold-plated TOSLINK plugs! 

[LapTop006]

Given that my current design uses a gigabit-rate fiber link that the customer uses with a plastic fiber that's about 200 meters long, and it works perfectly, I think that a plastic TOSLINK fiber that's 3 meters long will work without problem.

I'm impressed by that, wouldn't have expected it to work. Standard gig-E on multimode (glass) fiber doesn't have a distance rating much more than that.

[Daixiwen]

I have a 10 meter plastic toslink cable with absolutely no problem. And the good thing with plastic fiber is that when I wanted to shorten the cable a bit I could just use regular tools to cut the cable and put on a new plug. No need to be extremely careful about not braking the glass. The most "exotic" tool that I used was a razor blade to do a clean perpendicular cut on the fiber.

[tooki]

Yep. Heck, TOSLINK syncs even once the cable is halfway into the connector, it doesn’t even have to be seated properly. It’s very forgiving.

I’m actually kinda surprised we don’t use more low-cost optical fiber links using the same technology. Surely plastic optical fiber is cheaper than copper? And in some applications, it’d help eliminate ground loops that mess up audio recording. (I could imagine USB being done optically, for example.)

The only real downside is that it doesn’t carry power, but one could bond it to copper cable for that, perhaps even integrated into a single connector. That and minimum bend radiuses, though again it tends to be rather forgiving, and one could probably make it much thinner, to make the bend radius smaller.

[LapTop006]

I’m actually kinda surprised we don’t use more low-cost optical fiber links using the same technology. Surely plastic optical fiber is cheaper than copper? And in some applications, it’d help eliminate ground loops that mess up audio recording. (I could imagine USB being done optically, for example.)

The only real downside is that it doesn’t carry power, but one could bond it to copper cable for that, perhaps even integrated into a single connector. That and minimum bend radiuses, though again it tends to be rather forgiving, and one could probably make it much thinner, to make the bend radius smaller.

Corning do some USB optical cables although they integrate the transceivers.

[Richard Crowley]

At 3m (10ft) attenuation is NOT an issue.  TOSLINK was designed to be used with very inexpensve extruded plastic "cable".
Of course, as with ANY basically simple concept there are snake-oil shysters who will sell you something at 10x or 100x or 10000x the price with the completely baseless claim that it is "better".

The whole issue of "jitter" appears to be a relic of the early days of very simple consumer digital audio.  No modern digital audio uses the raw derived clock from the digital stream.  All modern consumer gear (and all modern commerical/broadcast, etc. gear uses local, crystal-referenced re-synchronization. 

Yes you will find audiophools even here in EEVblog Forum who defend the baloney. As with all audiophoolish concepts, we have never seen any objective proof of the myth.  Note that even the longest optical fiber connections (undersea cables) are thousands of miles/kilometers long in the harshest of conditions with very few booster/repeater amplifiers.  You could count the number on one hand.  Surely if "jitter" were a problem, you would not be able to read this message (or any of the other billions on the internet. 

Essentially, ALL of the "internet" is connected through undersea optical fiber.  Satellite communication has horrible latency because of the altitude of the geosynchronous Earth orbit. (around 500ms full duplex).  Anyone who has done live television via satellite knows that you must keep an offset program clock that is ~500ms ahead in order to hit your time slot properly.  (Been there, done that, have the T-shirt) You can see this every day on international news broadcasts where it takes an uncomfortably long time between when the local news-reader (in America "anchor") asks a question of the reporter out in the field on the other side of the planet replies.  You are seeing the combined latency in both directions.

Certainly, there are glass-fiber TOSLINK cables (and connector adapters from TOSLINK to proper glass-fiber).  They are commonly found in commercal operations where they use already-existing fiber cables, and/or where they are in large plants where they have to send the signal over distances too long to be reliably sent via dirt-cheap extruded plastic "cable".  But that is because of attenuation in the plastic waveguide and has nothing to do with "jitter".  If we can see absolutely flawless video which has been encoded into digital, send thousands of kilometers and reproduced on your TV screen (or smart-phone) with zero artifacts, why do we still believe in the myth of digital "jitter"?   

[Berni]

The same people have also made a optical Thunderbolt cable quite a while ago.

I seen it in a linus tech tips video where to prove a point they ran the cable from a computer to the next house where they had a thunderbolt hub to run a monitor and USB devices, all at full speed. Pretty impressive stuff.

[Richard Crowley]

The same people have also made a optical Thunderbolt cable quite a while ago.

Thunderbolt was originally conceived for optical interconnection.
In modern practice most shorter, less expensive Thunderbolt cables are copper because it is cheaper for short distances.

Ref: https://en.wikipedia.org/wiki/Thunderbolt_(interface)#Copper_vs._optical