I asked earlier this week why not use ethernet for computer interconnect
in contest operations.
I received 7 responses, mostly from active contesters, and one from a
technical type person. I categorize the "why nots" three ways:
1. cost. computers come with serial ports, don't come with
network cards, and network cards cost more
2. appropriate software doesn't exist
3. RFI
Most of the contesters indicated they thought it would be a good idea
and should solve some problems. (improved reliability, not all computers
that show up at multi-multi operations actually have enough ports). One
thought that the required bandwidth did not warrant use of network
cards.
Several disputed my assertion that the cost factor is in favor
of network cards, but you have to ignore my presumption of an upgrade
from DOS for this to be true. So non-DOS contesting software must
have enough advantages above existing products to justify yet another
increased cost over running DOS (newer OS needs more CPU, more memory,
more disk, etc.) But, unless it must interoperate with DOS cpus in
the same station, the non-DOS product can just as well use ethernet
as serial lines from a cost standpoint.
Interestingly, those that pointed out #2 were all in a
position to actually do something about it. I am too. But nobody
has sufficient motivation. If networking software for contesting
happens, it will likely be done by an eager volunteer--I doubt
anyone will figure a way to make any money at this.
And I quote the only person to indicate point #3:
>A few reasons:
>
>1. At 10 MHz signalling, Ethernet is a MUCH higher source of RFI.
>2. Although you probably wont notice it often (degraded performance
>unnoticed due to extreem over-kill from a bandwidth perspective, for
>this application), Ethernet is much more susceptable to RFI from
>transmitter local field radiation (neither RG-58, thin net, nor
>twisted pair -- even if shielded -- provides effective shielding).
>3. thin-net installations are very susceptable to disruption due to
>bad connector installation, cheap T-connectors, terminators being
>improperly installed -- a triped-over cable can bring the network
>down. Alternatively, 10-BASE-T requires extra cost hub, and in radio
>shack / contest / FD environment MUST use shielded twisted pair for
>satisfactory performance -- RJ45 connectors on most 10-BASE-T cards do
>not provide for shield connections; shielded Twisted-Pair cable s are
>not prevalent. No one uses Thick Ethernet installations anymore.
>1-BASE-5, Starlan, might be a good choice since lower signalling
>frequency offers more opportunity for RFI suppression.
>4. network OS and drivers are memory hogs = fewer QSOs in memory.
>3. serial interfaces are an order of magnitude cheaper and have more
>than adequate throughput.
>
>Use or discard above as you see fit.
>
>Regards,
>
>John Montague
>W0RUE
>ARRL Technical Advisor - Data Communications
>Vice Chair, IEEE 802, LAN MAN Standards Committee
Wayne, W5XD
w5xd@delphi.com
>From oo7@astro.as.utexas.edu (Derek Wills) Wed Jul 27 22:02:15 1994
From: oo7@astro.as.utexas.edu (Derek Wills) (Derek Wills)
Subject: SS Winners and mults
Message-ID: <9407272102.AA17585@astro.as.utexas.edu>
Tree said:
>>On a different subject, I delete messages having to do with grid
>>squares. It is just too different from how we do things now.
Paul WG0G replied:
>Do you take the same approach to new radios, antennas, and logging
>programs?
I understand what Tree feels. In international contests, the adrenaline
level goes up in CQWW when you hear your first JT or someone sending 599-40
for the first time, I don't think I could get excited about grid squares in
the same way.
If the squares are too small, it's too much like WPX, where the mults come
along with the QSOs anyway, and if they are too large we are back to CQWW
countries or zones.
I just can't see the post-contest comments - "So then this FG-12a calls
me" - "Wow!!"
Color me old-fashioned too.
Derek AA5BT, G3NMX
oo7@astro.as.utexas.edu
|