[comp.dcom.lans] Installing Thinwire Ethernet

milo@ndmath.UUCP (Greg Corson) (05/09/89)

I was thinking about a problem someone MUST have solved by now....if you
have a solution to this, let me know.

If you want to install thinwire Ethernet to a series of single-person offices
it seems to present a problem.  As I recall, you can't have a drop-cable
going from the T connector to the thinwire interface card.  This seems to
prevent you from having a simple wall jack that you can plug into to get
ethernet service (unless you imbed transcevers in the wall...which is an
extra cost considering a lot of computers already have them built-in)

So, how can you wire 10-20 offices so they can all easily tap into the 
thinwire?  The installation has to be neet...no big loops of cable hanging
down from the ceeling...etc.  All the offices would have to be on the same
segment of thinwire...sending a seperate thinwire cable to each one isn't
practical because there are TO MANY offices.

I suppose you could put in a wallplate with two BNC connectors and install
a short jumper when the outlet isn't in use.  When you connected a computer
you would run cable from one connector, to the T on the back of the computer
and back to the other connector on the wall plate.  I'm sure there must be
a better way than this though.  Anybody got any good ideas?

Greg Corson
19141 Summers Drive
South Bend, IN 46637
(219) 277-5306 
{pur-ee,rutgers,uunet}!iuvax!ndmath!milo
 

jjb@wsu-cs.uucp (J. Brewster) (05/10/89)

In article <1381@ndmath.UUCP> milo@ndmath.UUCP (Greg Corson) writes:
@>I was thinking about a problem someone MUST have solved by now....if you
@>have a solution to this, let me know.
@>[...]
@>So, how can you wire 10-20 offices so they can all easily tap into the 
@>thinwire?  The installation has to be neet...no big loops of cable hanging
@>down from the ceeling...etc.  All the offices would have to be on the same
@>segment of thinwire...sending a seperate thinwire cable to each one isn't
@>practical because there are TO MANY offices.
@>
Well, one approach you could consider is a multiport repeater.
I don't have any experience with the thin ethernet version, so I can't
tell you whether or not you have to terminate the ports when not in use.
The DEC version is called a DEMPR (sp?) and allows a star topology from
the device.  

-- 
J. Brewster               | "In this country, everything loose
jjb@cs.wayne.edu          | rolls to the West Coast."
...!mailrus!wsu-cs!jjb    | --Thomas A. Vanderslice, CEO of Apollo

kwe@bu-cs.BU.EDU (kwe@bu-it.bu.edu (Kent W. England)) (05/10/89)

In article <1381@ndmath.UUCP> milo@ndmath.UUCP (Greg Corson) writes:
>
>So, how can you wire 10-20 offices so they can all easily tap into the 
>thinwire?  The installation has to be neet...no big loops of cable hanging
>down from the ceeling...etc.  All the offices would have to be on the same
>segment of thinwire...sending a seperate thinwire cable to each one isn't
>practical because there are TO MANY offices.
>
>I suppose you could put in a wallplate with two BNC connectors and install
>a short jumper when the outlet isn't in use.  When you connected a computer
>you would run cable from one connector, to the T on the back of the computer
>and back to the other connector on the wall plate.  I'm sure there must be
>a better way than this though.  Anybody got any good ideas?
>
	Are you sure you don't work for AMP?  This question sounds
like the perfect set-up for telling you about AMP's new Thinnet Tap
System.  Some of you may have heard of it, but if not...

	AMP has designed a new cabling gizmo to alleviate some of the
problems of daisy-chaining thinnet thru walls among strings of offices
and labs.

	There is a wallplate with a single connector that brings a
dual coax out of the wall and terminates on a BNC.  This is just like
using two BNCs on the wall with two patch cords and a T, with an
important exception-  the AMP wallplate connector has shorting bars to
restore the net when you unplug the dual-coax patch cord.

	Unplugging the workstation from the BNC should not present a
problem, so there you have it:  all thinnet daisychaining problems
solved. 

	Well, not quite, in my opinion.  I feel obliged to tell you
that if you go daisy chaining to the max you will live to regret it.
But, perhaps, a moderate degree of daisy chaining could be tolerated
if this AMP gizmo really works as advertised.  I am a little nervous
about the shorting bars and the connector itself.  So, don't say I
didn't warn you and don't make the mistake of thinking that I am
saying that daisy chaining is OK.  I am not.

	But this is interesting enough that I would like to try a few
out and see how practical they are.

	Does anyone else have an opinion yet on this approach?

haas@wasatch.utah.edu (Walt Haas) (05/10/89)

In article <1381@ndmath.UUCP>, milo@ndmath.UUCP (Greg Corson) writes:
> ...how can you wire 10-20 offices so they can all easily tap into the 
> thinwire?  The installation has to be neet...no big loops of cable hanging

You can do this by installing the AMP Thinnet tap system.  Or, you
can bag thinnet and wait for the 10BASET standard to become final.

Cheers  -- Walt Haas    haas@cs.utah.edu    utah-cs!haas

sob@watson.bcm.tmc.edu (Stan Barber) (05/10/89)

We have used the AMP LAN-LINE system for thin Ethernet and it works well.
The installation instructions are clear and make it easy to get it right
the first time, especially if you have techs who have some experience at
regular BNC-type thin network installation. The installation is neat and
professional looking.

The problems are few, but are worth considering.

Your total run is still the same as it is for thin ethernet. Remember
that the lenght of the connector from the wall to device is twice its
length (dual coax in that cable). This means that the effective length
will be shorter AND you have to know where all the devices are likely to
be when designing things.

The connector once it is plugged into the wall outlet is hard to remove.
You need something like a nail file to release the connector. Some might
say this is a feature.




Stan           internet: sob@bcm.tmc.edu         Manager, Networking
Olan           uucp: {rutgers,mailrus}!bcm!sob   Information Technology
Barber         Opinions expressed are only mine. Baylor College of Medicine

brian@apt.UUCP (Brian Litzinger) (05/10/89)

From article <1381@ndmath.UUCP>, by milo@ndmath.UUCP (Greg Corson):

> If you want to install thinwire Ethernet to a series of single-person offices
> it seems to present a problem.  As I recall, you can't have a drop-cable
> going from the T connector to the thinwire interface card.  This seems to
> prevent you from having a simple wall jack that you can plug into to get
> ethernet service (unless you imbed transcevers in the wall...which is an
> extra cost considering a lot of computers already have them built-in)

A number of coax cable and connector manufacturers have a product
called triax, single shield, two wire coax.  I've seen it at the
local electronics/hardware store in my area.  There are also
connectors to convert triax to dual BNC.

Thus, you can run your network to triax connectors on your wall plates,
run triax to the computer in your office, and convert to BNC at the
computer.  The one draw back to this solution is that you have to
plug a shorted triax connector on the wall tap when it is not in use.

All in all though, seems like a fairly inexpensive solution, at least
until you find out how much triax costs. 8-)

<>  Brian Litzinger @ APT Technology Inc., San Jose, CA
<>  UUCP:  {apple,sun,pyramid}!daver!apt!brian    brian@apt.UUCP
<>  VOICE: 408 370 9077      FAX: 408 370 9291

tsmith@usna.MIL (Tim G. Smith ) (05/10/89)

In article <1381@ndmath.UUCP> milo@ndmath.UUCP (Greg Corson) writes:
>So, how can you wire 10-20 offices so they can all easily tap into the 
>thinwire?  The installation has to be neet...no big loops of cable hanging
>down from the ceeling...etc.  All the offices would have to be on the same
>segment of thinwire...sending a seperate thinwire cable to each one isn't
>practical because there are TO MANY offices.

I know you don't want to hear this BUT.....

The cost and praticality all depends on your viewpoint. The major cost
of running a single thin wire to each office always seems to be the
multi-port repeaters (I call them MPRs but I am told many folks call
them DEMPRs). MPRs are not cheap- BUT running around finding out who
is screwing up the network whenever some clown unplugs his pc the
wrong way also costs a lot in terms of wasted resources and pissed of
network managers.

Another thing to consider is whether offices ever change hands and if
so will it affect your subnets. [Office wars for the "best" offices
can wreak havoc with subnetting if office "moves" from one subnet to
another] If every office is "home runned" then moving an office from
one subnet to another is easy- if the offices are daisy chained then
you have to go run new wires.

Cabletron has a box called an MMAC that significantly reduces the cost
per port of using MPRs to star wire.

>I suppose you could put in a wallplate with two BNC connectors and install
>a short jumper when the outlet isn't in use.  When you connected a computer
>you would run cable from one connector, to the T on the back of the computer
>and back to the other connector on the wall plate.  I'm sure there must be
>a better way than this though.  Anybody got any good ideas?

Maybe the AMP wall outlets that some other folks mentioned will do the
trick for you if you absolutely, positively can't star wire.

good luck,
	Tim Smith	(formerly of the US Naval Academy-
					and still reading news there)
US mail:US Army, BRL				E-mail:
	SLCBR-SE				internet:tsmith@brl.mil
	Aberdeen Proving Grounds, MD 21005-5066	uucp	:...!uunet!brl!tsmith
MaBell :(301)278-6678 (or 6808)
Autovon: 298-6678
					

cfreese@super.ORG (Craig F. Reese) (05/12/89)

>>So, how can you wire 10-20 offices so they can all easily tap into the 
>>thinwire?  The installation has to be neet...no big loops of cable hanging
>>down from the ceeling...etc.  All the offices would have to be on the same
>>segment of thinwire...sending a seperate thinwire cable to each one isn't
>>practical because there are TO MANY offices.
>
>I know you don't want to hear this BUT.....
>
>The cost and praticality all depends on your viewpoint. The major cost
>of running a single thin wire to each office always seems to be the
>multi-port repeaters (I call them MPRs but I am told many folks call
>them DEMPRs). MPRs are not cheap- BUT running around finding out who
>is screwing up the network whenever some clown unplugs his pc the
>wrong way also costs a lot in terms of wasted resources and pissed of
>network managers.
>
I whole heartedly agree.  We just wired our new facility.  Our approach
was to put 2 BNC connectors/room.  Each of these ports runs to a wing 
computer room which contains a 19" rack.  In each rack there is a BNC
patch panel which presents the end of the station (office) wiring, and
the ports to the group of MR-9000s (Cabletron's MR) in the rack.  The MRs are 
connected into building wide backbones.  I believe the 
following issues guided our decision:

   1) the extra cable, connectors, and wallplate cost was minor
   2) all configuration (& thus control) exists in the racks 
      (lockable I might add).  We wired ~120 offices, and having
      that much "distributed" configuration would be hard to manage
   3) each machine is isolated from the others by the repeater
      (i.e. people can rearrange their offices without "killing" others)
   4) with the terrible lack of network management tools, having the
      machines 1 per link allow us to visually watch the network.  
      (the cabletron boxes have a few lights for each MR port.  This
      makes it easy to locate chattering PCs, and disconnected links)
   5) since we have two wallplates, we can run a star-loop if we
      want to.
   6) If we want to run more than one machine per office we can
   7) we can easily and quickly load balance across the multiple 
      thicknet backbones that the MR connect to by simply changing
      the patch panel.

The hardware costs for the network is higher due to the extra MRs, but
it has proved welcome in the past few weeks since we moved in and
set up shop...

craig (maybe this networking wasn't a waste of time afterall) reese.

*** The opinions expressed are my own and do not necessarily represent
*** those of any other land dwelling mammals....
-----------------
Craig F. Reese                           Email: cfreese@super.org
Institute for Defense Analyses/
Supercomputing Research Center
17100 Science Dr.
Bowie, MD  20715-4300

butzer@rugby.cis.ohio-state.edu (Dan Butzer) (05/12/89)

In article <9121@super.ORG> cfreese@super.org (Craig F. Reese) writes:
>So, how can you wire 10-20 offices so they can all easily tap into the 
>thinwire?  The installation has to be neet...no big loops of cable hanging
>down from the ceeling...etc. 

We use thin Ethernet to connect about 275 workstations to their respective
file servers (about 15 nodes per Ethernet).  We've had few problems 
daisy chaining workstations.  This includes installations within individual
offices, and in the student labs.  In the offices, we use plastic over the
wall conduit to hide the coax, and in the labs, we encase it in Panduit.
In an office, the incoming and outgoing coax ends just below the end of the
conduit (and has BNC connectors).  When you install a computer, you'll need
to add extensions. When an office does not need a workstation, 
then you use a barrel connector to patch the incoming, and outgoing 
coax together.

Depending on the nature of your users, this may not work.  If your users
like to connect and disconnect cables at will, and get into their hardware,
you might appreciate the isolation that the multiport repeaters provide.
(We also prefer Cabletron MR9000c's).  If you do decide to daisy chain
thin wire, you might want to consider investing in a GOOD time domain 
reflectometer.  When you do have a problem, you will want to find it very
fast because it will take several people down.  Also, you'll want to be
extremely picky about the quality of the cables you use.  Dont even consider
solderless connectors, or any cable other than RG58 A/U.  (NOT RG58 /U !!!)






-=-
______________________________________________________________________________
Dan Butzer - The Ohio State University, Dept of Computer & Information Science 
             Net:   butzer@cis.ohio-state.edu     Voice: (614)292-7350
             Snail: 2036 Neil Ave Mall, Room 245, Columbus, Ohio, 43210

ron@ron.rutgers.edu (Ron Natalie) (05/13/89)

Most of the available multiple coax stuff (dual-RG59, twinax, triax) are at
impedances other than 50 Ohms.  Twinax is like 150 ohms.  This doesn't
work at all on Ethernet.

-Ron

hubey@pilot.njin.net (Hubey) (05/13/89)

In article <179@usna.MIL> tsmith@usna.MIL (Tim G. Smith ) writes:

> In article <1381@ndmath.UUCP> milo@ndmath.UUCP (Greg Corson) writes:
> >So, how can you wire 10-20 offices so they can all easily tap into the 
> >thinwire?  The installation has to be neet...no big loops of cable hanging

> 
> The cost and praticality all depends on your viewpoint. The major cost
> of running a single thin wire to each office always seems to be the
> multi-port repeaters (I call them MPRs but I am told many folks call
> them DEMPRs). MPRs are not cheap-

How expensive is NOT CHEAP??  How many  ports ?? Can you please
provide  adress or tel # ??

We are/might be doing things along similar lines and we need some idea
as to cost etc.


> Cabletron has a box called an MMAC that significantly reduces the cost
> per port of using MPRs to star wire.

CAn you please provide more data ?? on Cabletron MMAC etc. Thank you.


> >I suppose you could put in a wallplate with two BNC connectors and install
> >a short jumper when the outlet isn't in use.  When you connected a computer
> >you would run cable from one connector, to the T on the back of the computer
> >and back to the other connector on the wall plate.  I'm sure there must be
> >a better way than this though.  Anybody got any good ideas?
> Maybe the AMP wall outlets that some other folks mentioned will do the
> trick for you if you absolutely, positively can't star wire.

Are there any gadgets that will automatically provide the terminator
impedance when a  workstation is disconnected ??


> good luck,
> 	Tim Smith	(formerly of the US Naval Academy-
> 					and still reading news there)
> US mail:US Army, BRL				E-mail:
> 	SLCBR-SE				internet:tsmith@brl.mil
> 	Aberdeen Proving Grounds, MD 21005-5066	uucp	:...!uunet!brl!tsmith
> MaBell :(301)278-6678 (or 6808)
> Autovon: 298-6678
> 					

Mark 
-- 

 hubey@OSultrix.montclair.edu       	hubey@pilot.njin.net
 hubey@apollo.montclair.edu 		VOICE:  201-893-5269                   

jeff@nsipo.arc.nasa.gov (Jeff Burgan) (05/18/89)

>Well, one approach you could consider is a multiport repeater.
>I don't have any experience with the thin ethernet version, so I can't
>tell you whether or not you have to terminate the ports when not in use.
>The DEC version is called a DEMPR (sp?) and allows a star topology from
>the device.  
>
>-- 

You don't have to terminate the unused ports on a DEMPR. DEC also
has a box called a DESPR (i.e. single port as apposed to multiport repeater)
On one segment from a port on a DESPR or DEMPR, you are allowed
29 stations and the cable length can not exceed 185 meters. 

Jeff

howell@ecsvax.UUCP (Doc A. Howell) (05/18/89)

  Dec has a product that may be the answer to drop cables drom a 
thinwire wallplate. It is called, Thin Wire Ethernet Daisy-Chain connector.
Catchy huh? Supposedly, this device/wallplate/concept allows up to 8
wallplates to be daisy chained on a single thinwire segment from which
a single BNC cable may be run between the workstation and the wallplate.
The Tee is behind the wallplate thus allowing the single coax. This 
drop cable can be up to 2.5 meters (8.2 feet) from the wallplate. All
of this information is rather poorly described in the DECconnect System
Planning and Configuration Guide. 

  Way back several months ago, this same question of drop cables arose.
Several people said they were running short drop cables from the wallplate
with the Tee behind the wallplate. IT IS AGAINST THE SPECS OF ETHERNET
TO DO THIS. They said it worked. I don't know but it is realistic to
expect that you can get away with certain violations as long as you 
compensate by enforcing stricter limitations in other areas such as 
Max distance, and MAx stations per segment (note the 8 above).

  I will probably try this sometime in the future, I may be forced 
to. DEMPR's are not cheap, neither are MMACs, Multiconnects, or any
of the other similar devices that break thinwire off of an ethernet.
I am sure DEC's wallplate isn't cheap either. 

  Will those people that claimed to be doing this several months ago
like to follow up on the subject again? Has anyone tried DEC's new
product? 

morrison@norby.acns.nwu.edu (Vance Morrison) (05/19/89)

Hello,


I realize that this does not directly the question about Tee's on
thin ethernet but I thought people would be interested in it.

A novel ethernet product I have seen reciently is twisted pair ethernet with
Thin net converters.

The idea is this, you run twisted pair ethernet from your phone closet to
wall plates in each room.  Each room has a Twisted pair converter mounted in
the wall plate.  Then you just daisy-chain everything (up to 10) in the room 
with thin net and connect it to the wall plate.  

This I believe is a very nice solution.  

Costs

	Twisted pair Hub (8 Ports)	$2000 
	Twisted pair/Thinnet converter	$450

Thus the cost per room is about $750.  But if there are 3 machines in the
room this is a $250/machine, not bad.  

I have seen these things advertized by Cabletron, but I am sure there are
other vendors selling them

If there is interest, I mail/post specifics


Vance

tony@cc.brunel.ac.uk (Tony Begg) (05/22/89)

I looked for a solution that would not involve extra cable length for
unoccupied offices, and use existing BNC metalware in a cost-effective way.
There just didn't seem to be any appropriate wallplates.  Maybe this AMP
LAN-LINE thing is what I've been looking for but it sounds a bit complex.

In the absence of any neat solution, I have been installing thin ethernet with
a barrel connector in each office, which is broken to connect to equipment,
a second barrel connector used, with standard made up BNC cables and a T
piece at the back of the workstation or whatever.  All that is wrong with
this approach (if you ignore breaking the network to add equipment which a
lot of other solutions share) is it is not particularly neat or mechanically
sound.  I believe it could be made so with a two part injection moulded wall
plate, in the shape of a truncated equilateral triangle (ie sort of hexagonal).

One part screws to the wall and the other part is a cover which is removed
to add or remove equipment.  The BNC plugs with their barrel connector
locate in channels along the sides of the triangle.  An "empty" office has
the single plug-barrel-plug located along the base of the triangle.  To connect
equipment, the barrel is broken, a second barrel added to the free plug, and
these connected to the two device cables, then the two plug-barrel-plug
assemblies located along the other two sides of the triangle, so the twin
cable to the device exits from the "top" of the triangle.  Location would be
in tubular channels in the part of the wall plate screwed to the wall, with
longitudinal restraint by protrusions into the channels that fit the bit of
the BNC plugs with a smaller diameter (where you can see the slot for the
bayonet pin).  I noticed that the length of the various barrel connections was
about constant whereas the diameter of the centre of the barrels varied between
manufacturers, so that this method of restraint is more metalware independent 
than a protrusion between the two plugs.  The triangle would need to be about
5.75 ins along a side, and truncated to give a hexagon with three opposite sides
about 3.25 ins (along which the channels for plug-barrel-plug run) and the other
three sides 1.25 ins (where cables enter and leave).  The wall plate is 3
fold symmetric.  I am trying to convince our Materials Tech department to make
the injection moulding (or a cheaper vacuum formed version).

Of course for the rectilinear minded as long as the channels in the wall
plate formed a triangle, the wall plate could be about 4 ins square.