bandwidth limitation on optical cabling? (1 Viewer)

[Ted Lee]

hi all -

there is a thread going on in basics about the old dead horse debate: optical vs. coax. around post 30 a htf'r mentions that optical has less bandwidth.

Toslink has a potential 6MHz bandwidth
Coax has a potential 500MHz bandwidth

i'm not really up on this stuff, so i was wondering what everyone's thoughts were about this?

if possible, i don't really wanna get into the coax vs. optical debate. i'm looking more for thoughts on this bandwidth concept.

thx,

ted

 

[Lew Crippen]

Ted, I don’t know the specs of ‘Toslink’, but I’m intrigued by three things: the use of the word ‘potential’, the almost two orders of magnitude difference in the numbers provided and the metric (MHz) being used.

To begin, I’m skeptical of why ‘potential’ is cited, rather than the performance that is actually being delivered.

But aside from that quibble, I am interested to know why the optical link would be so restricted compared to copper. After all, serious commercial applications requiring considerable bandwidth are almost exclusively fiber, not copper, so I am highly suspicious of these numbers. If copper is almost 100 times more efficient than fiber in this very limited sense, why is it not being used more commercially?

Finally, although audiophiles love to discuss frequency and bandwidth in terms of Hertz, this is an analog concept. The correct metric for digital metrics is ‘bit’, kilobit, megabit, etc. This is because digital transmission depends on ‘0’s and ‘1’s following each other in succession down the pipe.

Taking all of these things into consideration, I’d need a technical explanation of some detail in order to convince me that these numbers are even close to being accurate.

 

[Ted Lee]

lew - i'm glad you chimed in. i know you know your stuff.

that's pretty much what i was thinking. those specs just didn't "feel right", but i certainly couldn't explain why.

for me, in a nutshell, if optical was so *significantly* inferior, it wouldn't be as prevalent as it is.

 

[David_Stein]

if you think about computers, it makes perfect sense to talk about digital transmission in terms of Mhz.

i dont know for sure if toslink is exactly like that, but i am assuming it can only transmit one bit (light or no light) at a time, though potentially with optical fiber you can have as many strands as you want in a cable and trasmit as much as you like.

then you have to think (briefly) about how fast that the information can be sent down the channel. optics move at the speed of light (i believe, it might be slightly slower i suppose, or at least apparently if it has to reflect alot giving it a longer pathlength but thats all quantuum that makes my brain hurt so i am going ignore it while in the copper it moves as fast as electrons can in a metal. basically they are both going to be fast enough to eliminate that as limiting factor (i would assume).

the last thing to consider is how fast the device at the ends of the cable can switch between states. both at the recieving end (though i would assume that it is passive and switches a lot faster than the sending end) and the sending end.

im working on basically no knowledge of the digital coaxial and toslink specs, but if i had to guess, this is where i would guess the bottleneck is. optical technology is a lot younger than electrical technology and i would guess that either through the toslink specs or physical limitations, the optical signal generation device that switches the signal to an wave packet for transmission in the cable is limited to somewhere around 6Mhz, meaning that it can only switch 6 million times a second, hence sending 6 million bits of information accross the line per second. once again, if i had to guess, i would be willing to bet its in the toslink specs. im guessing 6Mhz is more than enough for any audio application and they restricted it to save money for manufacturers using this spec.

if i am wrong, im sorry, im just working from first principles and i would love to be corrected.

 

[Jagan Seshadri]

I would suspect that the quality of the optical material used, and the quality of the outer cladding (and therefore the uniformity of the critical angle over the length of the fiber) would play a large role in overall bandwidth capacity.

Think of it as a telescope made with finely-machined mirrors and lenses able to resolve fine detail (i.e. high bandwidth) versus a similar telescope using drug-store-quality mirrors and somewhat cloudy plastic lenses. You'd be able to resolve a lot less detail (i.e. lower bandwidth).

Optical has the potential for much higher bandwidth than coaxial cable, but it has to be *good* optical.

-JNS

 

[Nick V]

My understanding was that Fiber-optic cables CAN carry much more information than they currently do on consumer electronics, but the limiting factor here is the interface, not the cable itself.

The reality is that on consumer electronics, coax can carry substantially more bandwidth than optical.

 

[Ted Lee]

The reality is that on consumer electronics, coax can carry substantially more bandwidth than optical.

okay, let's assume that is true.

then the next question would be, what is the maximum bandwidth (or bit) requirement for home audio? is the optical good enough to "cover" all the bits? is the coax simply above and beyond (kinda like driving a porsche when a vw will work?)

thx for the replies so far guys...

 

[David_Stein]

>>I would suspect that the quality of the optical material used, and the quality of the outer cladding (and therefore the uniformity of the critical angle over the length of the fiber) would play a large role in overall bandwidth capacity.

Think of it as a telescope made with finely-machined mirrors and lenses able to resolve fine detail (i.e. high bandwidth) versus a similar telescope using drug-store-quality mirrors and somewhat cloudy plastic lenses. You'd be able to resolve a lot less detail (i.e. lower bandwidth).

Optical has the potential for much higher bandwidth than coaxial cable, but it has to be *good* optical.

 

[David_Stein]

>>then the next question would be, what is the maximum bandwidth (or bit) requirement for home audio? is the optical good enough to "cover" all the bits? is the coax simply above and beyond (kinda like driving a porsche when a vw will work?)

 

[John Royster]

Hi. I just so happen to work with fiber every day so...

Toslink is indeed limited to about 6 Mhz. I don't remember the exact figure but that sounds right. The fiber core is so big that you get tons of loss and internal reflections. So as the signal gets faster and faster the tolerances for that loss and relections get smaller and smaller.

Toslink as a fiberoptic medium is really very poor. You can now transmit on fiber optic cable in the 100s of Gigahertz.

Hope that helps. Also the specs from toshiba set this limit because all fiberoptic transmissions are a factor of three things - the transmitter itself, the physical cable (and this includes any junctions like the cable to transmitter junction), and the receiver.

Another tid bit is toslink is limited to 5 meters or 10 meters depending on which toslink transmitter is used. In real fiber optics the distance is 50-100 kilometers.

link to toslink specs on different transmitters/receivers:
http://www.toshiba.com/taec/cgi-bin/...ategoryID=7071

 

[Chu Gai]

According to AES preprint 4826, any cable which can reliably pass a signal and allow the receiving unit to lock on to it is satisfactory. Certainly at the present time given the equipment in existence, both approaches are capable of meeting this objective. Whatever the bandwidth differences may be of the two cables is not being challenged by audio or video equipment.

As most know, toslink is available in plastic and glass (and of course varients within those two general classes). There is signficantly more signal loss in plastic but this is only an issue with very long lengths. This fact is recognized by the AES which has a recommendation, presently under review, advising that for lengths over 5 meters, glass is the preferred medium in professional applications.

Worth consideration is that since toslink doesn't create or emit EMI, there is no need for manufactures to add circuitry which can limit bandwidth in the equipment. Constrast that with say SP/DIF where this is a not unknown issue. A typical fix for eliminating EMI/RFI (FCC you know!) is to use transformers which in and of themselves, constrain bandwidth. This would suggest that to properly answer the question, one needs to consider the system as a whole. It's like asking which car is faster: one with an 8 cylinder 5 liter engine or one with a 4 cylinder 1.8 liter engine? If the former has a governor on it limiting its maximum speed to 85 mph it's the little POS 1.8 liter. I think one needs to look at the entire picture...IMHO of course.

 

[Kevin C Brown]

FWIW, I have seen the bandwidth differences between optical and coax mentioned in Stereophile as well. I remember them saying something about "digital fog" for Toslink. They always default to the coax connection.

I also have a suspicion that optical is probably worse in terms of jitter too, in that you have an extra electrical-to-optical and back conversions. Theoretically, optical is a better medium for jitter, because coax can be affected by RF/EMI, but just because of those extra conversions, optical is more than likely to be worse.

 

[Chu Gai]

Only if the lengths are exceedingly long Kevin would the jitter contribution of the cable be a potential factor.

 

[David_Stein]

how can it be possible for a digital signal to be jittered? no matter what that digital signal still has to be converted back into a analog signal once its recieved, before that its just 1s and 0s without a time component. at first i thought i could be wrong about this, and that the signal could just be converted on the fly with whatever time anomolies were introduced in the first conversion problem, but then i realized that to decode any one sample, you have to wait for 16bits or 24bits or some set of bits to pass (reguardless of time). so any jitter in that digital signal between each bit in a set would be eliminated right there. after that, and to deal with the fact that they are sending 6 or 7 channels of audio, they would have to reclock, right? please say thats right, because im already annoyed at the lack of reclocking that is done.

 

[Lew Crippen]

if you think about computers, it makes perfect sense to talk about digital transmission in terms of Mhz.

Could you be a bit more specific David? The term Mhz in computers is usually used to refer to the clock speed of the CPU. Data transmission is normally cited in terms of kilobits—for example you might have a 56 Kb line or modem; not a 56 MHz connection.

 

[Lew Crippen]

dts 24/96 would be the highest bitrate, right? times 6 channels, maybe 7 if you could be 6.1. so 24 bits*96*1024(samples/second)*7 channels would be the number. i would guess that the answer is 17Mhz if it werent compressed. now i dont even know if 6.1 dts 24/96 even exists. or if dts is actually compressed or if its just a lossless compression scheme.

I don’t know the data transmission scheme in use for digital audio, but there are two general possibilities: each channel can be transmitted in parallel (as suggested above) or the transmission is serial in nature where the channels are intermixed in some fashion and identifying information supplied by the transmitter is decoded by the receiver and the single bit stream is decoded into 6 (or 7 or 8) discrete channels. I expect that the transmission is serial, due to the problems in timing parallel bit streams (While this is probably not an issue over 5 meters, it is a considerable issue for OTA, cable and satellite DD transmissions. Parallel transmission here would mean that your satellite receiver (a Hughes E86, for example) would have to be capable of receiving the 6 parallel bit streams and retiming them before sending them on the way. Given that data needs to be stored for some nanoseconds for this to occur, as well as several other issues to resolve, it is not likely that these, receivers have this level of sophistication—especially as I have seen a non-HD, but DD receiver sold for around $100)

This strongly argues for a single, serial transmission of 1’s and 0’s. In this case the media only needs to transmit light or electrical impulses fast enough for a single bit stream to contain enough information for all 6 channels. And I would strongly suspect, that even if the transmission occurs in parallel, the nature of optical fiber requires that there is a separate, discrete path for each signal. It follows that even the plastic fiber media would (in this case) have to have the ability to carry each light stream separately.

This again results in the constraining factor being how fast light can be transmitted via the fiber link. So long as it is fast enough to handle all of the 1s and 0s, nothing else makes any difference. If, for example the fiber can only transmit 1s and 0s fast enough to decode eight channels and digital coax can transmit enough data to decode 800 channels, there can be no additional benefit.

QED

The bandwidth argument not only uses improper terminology, but it is misleading and in the end, not logical.

None of the above reasoning should be construed as to my having a preference for either digital coax or optical. I have none, although I find the connection argument intriguing, although not convincing, as most people don’t jiggle or jar their A/V gear very often. : These connections don’t need to withstand anything much more than the occasional cat jumping up on some gear.

 

[David_Stein]

>>Could you be a bit more specific David? The term Mhz in computers is usually used to refer to the clock speed of the CPU. Data transmission is normally cited in terms of kilobits—for example you might have a 56 Kb line or modem; not a 56 MHz connection.

 

[Lew Crippen]

DTS is compressed.

AFAIK, only PCM is lossless.

 

[David_Stein]

thanks. i think im going to go see if i cant figure out how compressed it is on average.

also, i jsut want to point out again that i am not real knowledgable on either of the two digital audio cable specs, i only have general knowledge on technology and what its capable of, so despite my long winded replies i am very apt to be incorrect and would like, for my knowledge and the knowledge of others, to be corrected.

 

[David_Stein]

from: http://www.hifi-writer.com/he/dolbydts/dolbydts.htm

Full backward compatibility with all existing decoders. (Existing decoders will output a 48 kHz signal.)

so it seems that full regular non-24/96 dts is 48kHz and takes up to 1536kb/s. and it is, I believe, 24bit for full rate and i believe allows for 6 channels (5.1). so if we just multiply that by 2 (to go from 48k samples per second to 96kHz) and by 7/6 (to go from 5.1 to 6.1) and you still end up with around 3.6Mbps, or 3.6Mhz needed. once again, this is all unverified (though i tried as best i can).