Jack Carroll wrote:
On Fri, Apr 07, 2006 at 01:37:11PM +0930, Tom Cook wrote:
On 4/7/06, Jack Carroll <[EMAIL PROTECTED]> wrote:
On Fri, Apr 07, 2006 at 12:37:41PM +0930, Tom Cook wrote:
For 1, low latency is critical. You want as little delay between the
sound
hitting the mic and it being returned to the speakers as possible. If
the
delay becomes noticeable to the ear, you're in trouble. Also any
processing
you do in the loop adds to the delay, so you can't eat the whole delay
budget in your digitizer. Ethernet is simply not good enough
I'm not so sure about that. TCP/IP maybe isn't good enough, but
that's not the only low-level protocol that Ethernet can carry. A 480-bit
frame (10 audio channels with 32-bit data, ECC, and 20% overhead) would
take
4.8 uS to transmit. That's about the time it takes sound to travel 1/16"
through air, or about 1.5 mm.
Until there is other traffic on the network...
No other traffic can be permitted on the network. For that matter,
TCP/IP can't be permitted on the network, and CSMA/CD is absolutely
incompatible with audio applications. This has to be a dedicated network in
order to ensure absolute predictability for every time slot. Think of it as
T3 over Cat 6 twisted pair. Only the hardware layer of Ethernet would be
used, and that's only to make use of off-the-shelf hardware that has the
data rate and distance needed in professional recording applications.
If Ethernet hardware weren't already so common, this kind of thing
would probably be done with coax for improved EMI immunity, because 10Base2
or 10Base5 doesn't begin to approach the bandwidth RG-58 can deliver over a
100 meter run.
You can still use coax for Ethernet, you just need to add the cable
drivers. However, the coax is probably more subject to damage.
--
JRT
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
