Hello Timothy,

"Timothy A. Seufert" wrote:

I believe his point was that one does not need highly or even moderately
expensive cable to get excellent performance with well designed digital
transmission systems. CAT-5 bears his point out quite well.


I concur that that was the point Steve was trying to convey. But in a
digital transmission system operating at 100Mbps with cable runs of
100's of feet, I don't believe that the excellent performance observed
can be ascribed exclusively to the cable (although the cable does need
to be spec'd to handle this data rate-distance). Receivers,
transmitters, and error correction/control features associated with
transmission protocols (link, network and transport layers) also play
a critical role. My point is that a TOSlink-based transmission system
is nowhere near on the same level as what one might find in a
CAT-5/100Base-T installation.

BTW, what does TCP/IP have to do with Ethernet? TCP/IP is a higher
layer protocol that couldn't care less whether its packets are
transmitted over Ethernet or some other physical layer. You seem
confused about the delineations between different networking layers as
you consistently talk about TCP/IP when, for the purposes of this
discussion, you should be concerned only with the physical layer.


I agree that I made some fundamental (but reasonable) assumptions
here, but again my point was, and is, that focusing exclusively on the
physical layer is missing the larger picture of what's really going
on. But even on the physical level, I don't think anyone wants to
convey the suggestion that, by analogy, you'll have no transmission
errors with a 100 foot TOSlink cable. I don't believe Steve intended
to suggest this, but it is possible that someone could interpret his
statement that way, hence my original post.

b) TOSlink-based interconnections are based on a one-way,
point-to-point S/PDIF protocol, and not a two-way, network TCP/IP
protocol. S/PDIF uses a single parity bit for error correction.
Ethernet/TCP/IP provides for multi-bit parity checking, checksums, and
cyclic redundancy checks and supports requests for retransmission of
data packets to minimize transmission errors. S/PDIF is nowhere near
as robust as TCP/IP.

So what? In practice, even though Ethernet was engineered with
unreliable transmission as an acceptable tradeoff, errors are very rare
indeed on 100Base-T Ethernet. You can manage to get a noisy link, but
typically this is due to using crap network cards or poorly made cables
or violating maximum segment length, etc. It costs very little to build
Ethernet links that seldom lose a single bit.


Here you are supporting my argument. To get it to work as intended,
you need to implement it correctly. I'm not convinced that the
typical consumer will always follow best practices when mixing and
matching equipment and interconnects. This is why I questioned
Steve's argument that inferred that the characteristics of a
CAT-5/100Mbps implementation apply equally to a TOSlink
implementation. I don't consider that the best argument to make his
point because of the differences I cited.

c) TOSlink cable will be called upon to transmit a maximum data rate
of 12.8 MHz, limited by the transmitter and receiver used; most
transmitter/receiver components you are likely to find in audio
equipment are spec'd at a lower 6 MHz data rate.

Low bandwidth requirements mean that the cable doesn't have to be very
good to work perfectly, once again reinforcing the original poster's
point...


Actually, he didn't make this point. I was trying to make it for him.


I've not seen any
maximum transmission rate specs for TOSlink plastic fiber cables, but
the Belden site
(http://bwcecom.belden.com/college/techpprs/wcfsbetp.HTM) suggests it
is "limited to low bandwidth of a few megahertz".

Literally right after it says: "The effect is that the signal can only
go a few feet, maybe 20 or 30 feet." This suggests that to the author a
"few megahertz" could mean 20 or 30 MHz.


It is not at all clear, but in the worst case, if the bandwidth is
only say 3MHz over 20 to 30 foot distance, then to get a minimum 6MHz
bandwidth, one should limit TOSlink cable length to 10 to 15 feet.
This is in line with the AES recommendations. For 96kHz or 192kHz
signals, and/or more than 2 channels, then we're likely talking a
12MHz transmitter system in which case TOSlink cable length should be
limited to 5 to 7.5 feet. So again, if you keep cable lengths short,
things should be fine ... but I'd still really like to see a date
rate-distance spec for a TOSlink cable.

BTW, I'm not anti-TOSlink in any way, shape or form.

Best regards,

Terry