"Larry W4CSC" wrote in message
...

Ethernet requires a router and ethernet hubs to
connect devices.

No it does not. Ethernet over 10base-T (cat5 cable) requires a hub and a
cable to every device. The now almost obsolete thin ether net (10base-2, or
coax) would allow you to run a cable from device to device, using BNC T's at
every device.

USB, on the other hand, WOULD let the GPS talk directly to the chart
plotter.

No. USB works with one master and many slaves. Generally the computer is the
master an all other devices are slave. To make a GPS master, it would
require different USB hardware inside the GPS and quite some computing power
to behave as a USB master.

But, USB wouldn't work well in a broadcast situation because
it only allows two devices to talk to each other.

USB is master-slave. So only the master can initiate communications to a
slave, by asking if the slave has something to say. Slave can NEVER talk to
eachother.

NMEA2000 (CAN based) isn't all that bad, the problem is that it is not an
open protocol and you have to pay heavily to get your first NMEA2000
compliant device on the market. Buying the standard documents, test suite,
manufacturet and product ID for the first product costs about $10,000!

Like I have mentioned before, NMEA-0183 could well be upgraded to higher
speeds and a bidirectional bus (RS-485). Something like combining NMEA
(point to point, but RS-422) and Seatalk (broadcast but single wire) into
high speed RS-485. Still cheap to implement.

Wireless, either 802.11-something ethernet or Bluetooth is the best
answer. Wireless uses no wireway space. Instruments can be placed
anywhere you can get DC to them.

If you can get DC to an instrument, you can also get a twisted wire to that
instrument.

Wireless is too unreliable. When I walk away from my Bluetooth multiplexer
with my Palm in hand, I lose conact after one brick wall and 5 meters
distance. Even my WLAN stops at two concrete floors.

So imagine what happens in a metal hull.....

Meindert