timr@uqcspe.OZ (Tim Roper) (11/24/86)
Is there a cheap (reliable) method of connecting a workstation with a thin ethernet interface to an existing thick ethernet coax? I know about the DEC DEMPR* which looks good if you have many thin stations. We are only looking at a couple of thin stations at the moment and so would be tempted by a cheaper method. Of course it is recognised that numbers of thin stations may increase so will have to look at the cut off point between cheaper methods (if they exist) and the DEMPR method. [If this topic has been well covered before a reference would be appreciated.] -Tim. * The DEMPR connects up to 8 thin ethernet segments to a thick ethernet via a normal transceiver.
dave@pta.OZ (Dave Horsfall) (12/01/86)
In article <1097@uqcspe.OZ>, timr@uqcspe.OZ (Tim Roper) writes: > Is there a cheap (reliable) method of connecting a workstation with a > thin ethernet interface to an existing thick ethernet coax? Unless I've totally missed the point, all you need is a simple little adaptor - BNC-to-N type, and of course a BNC T fitting. Such adaptors are available from most electronic suppliers. Dave Horsfall Sun Computer Australia
eriks@yetti.UUCP (Eriks Rugelis) (12/04/86)
In article <157@pta.OZ> dave@pta.OZ (Dave Horsfall) writes: >In article <1097@uqcspe.OZ>, timr@uqcspe.OZ (Tim Roper) writes: >> Is there a cheap (reliable) method of connecting a workstation with a >> thin ethernet interface to an existing thick ethernet coax? > >Unless I've totally missed the point, all you need is a simple >little adaptor - BNC-to-N type, and of course a BNC T fitting. >Such adaptors are available from most electronic suppliers. > >Dave Horsfall >Sun Computer Australia i've heard others promoting a BNC-to-N connection, butting thick and thin coax... but i've never heard of anyone actually DOING it! don't you introduce the risk of reflections at the interface? (would i be able to hear a ringing on the cable if i got out my stethescope? :-) ) but seriously... my physics isn't up to snuff in this regard... does anyone out there use this in a live setup? would my favorite ethernet parts vendor support a network configured in this manner? -- Voice: Eriks Rugelis Ma Bell: 416/736-5257 x.2688 Electronic: {allegra|decvax|ihnp4|linus}!utzoo!yetti!eriks.UUCP or eriks@yulibra.BITNET
jc@piaget.UUCP (John Cornelius) (12/05/86)
If the characteristic impedances of the two cables are identical and the BNC/N adapter does not introduce a lump impedance different than the characteristic impedance, the simple transition connector will work. I note that BNC hardware comes in 50 ohm and 75 ohm characteristic impedances. -- John Cornelius (...!sdcsvax!piaget!jc)
lindberg@chalmers.UUCP (Gunnar Lindberg) (12/05/86)
In article <1097@uqcspe.OZ> timr@uqcspe.OZ (Tim Roper) writes: >Is there a cheap (reliable) method of connecting a workstation with a >thin ethernet interface to an existing thick ethernet coax? The simplest way is to just use N-to-BNC connectors and connect the piece of thin cable to the thick one. A BNC-Ethernet-terminator is easily made from four 200 ohm, 1/2 W resitors put in parallell. We've run with a mixture of 20 m thick and 100 m thin for quite a while with absolutely no problems! However: + You must scale the max lengths (500 m Thick = 185 m Thin). + If someone disconnects a BNC-connector somewhere, the whole net is broken. Active devices, e.g. a DEMPR prevents this. Good Luck, Gunnar Lindberg (lindberg@chalmers.{UUCP,CSNET,SUNET}) Department of Computer Science Chalmers University of Technology S-412 96 Gothenburg, SWEDEN
brian@sdcsvax.UCSD.EDU (Brian Kantor) (12/06/86)
We've tried connecting thin and thick ether cables together with lots of adaptors and not had any noticeable problems. Using a TDR (Time-Domain Reflectometer) to look at the mixed cable I saw some small impedence lumps (discontinuities) but they are on the same order as those I see where transceivers are tapped in, so it seems to me that there is little or nothing to worry about. We're using garden variety RG58/U cable with crimpon BNC connectors, ordinary N to BNC adaptors, and N barrel fittings. Nothing special. Brian Kantor UCSD Office of Academic Computing Academic Network Operations Group UCSD B-028, La Jolla, CA 92093 USA
gdw@sslvax.UUCP (12/09/86)
In article <447@yetti.UUCP> eriks@yetti.UUCP (Eriks Rugelis) writes: >i've heard others promoting a BNC-to-N connection, butting thick and >thin coax... but i've never heard of anyone actually DOING it! > >does anyone out there use this in a live setup? We currently have a VAX 750 connected to thick ethernet via an H4000, this is connected to a thin ethernet network of 20 Suns using a BNC-to-N connector -- no problems! >would my favorite >ethernet parts vendor support a network configured in this manner? There's no reason why it shouldn't work, so why not? -- +----------------------------+----------------------------------+ | Grenville Whelan, | Tel - +44 625 29241 | | Software Sciences Ltd, | | | London & Manchester House, | | | Park Street, | EMAIL - gdw@ssl-macc.co.uk | | Macclesfield, UK. | UUCP - ...!mcvax!ukc!sslvax!gdw | +----------------------------+----------------------------------+ "Me, i've just died but some machines keep on humming..."
jqj@gvax.UUCP (12/10/86)
I have a similar problem that is not solved by the proposed N-to-BNC connector approach: I have thick Ether in raceways in the hallways, with arrangements for drop cables (xcver cables, that is) into individual offices. Now people want to use workstations in their offices that have only thin-Ether transceivers builtin. I gather that if I put a BNC tee in my thick Ethernet the tail of the T must be quite short, so the thin-Ether transceiver (built into the back of the workstation) must be located very near the thick Ethernet cable -- i.e. in the hallway. Having workstations located in hallways near the ceiling is not appealing in our environment. Any (serious) suggestions on what to do?
jc@piaget.UUCP (John Cornelius) (12/12/86)
In article <650@gvax.cs.cornell.edu> jqj@gvax.cs.cornell.edu (J Q Johnson) writes: >I have a similar problem that is not solved by the proposed N-to-BNC >connector approach: I have thick Ether in raceways in the hallways, with >arrangements for drop cables (xcver cables, that is) into individual >offices. Now people want to use workstations in their offices that have >only thin-Ether transceivers builtin. I gather that if I put a BNC tee >in my thick Ethernet the tail of the T must be quite short, so the thin-Ether >transceiver (built into the back of the workstation) must be located very >near the thick Ethernet cable -- i.e. in the hallway. > >Having workstations located in hallways near the ceiling is not appealing >in our environment. Any (serious) suggestions on what to do? There should be little or no penalty for interrupting your thick cable and running a loop of thin cable to the workstations. What may be most appealing might be to leave the thick cable as a backbone and then have a small number of fairly long local loops as in: =======| |==============| |============| |=== | | | | | | X X X X X X | | | | | | X X X X X X | | | | | | --X--- --X--- --X--- Where each X is a workstation. This will require cutting the thick cable and adding N style connectors to the new ends and then using the N to BNC converters to initiate the thin local loops. Earlier discussions in this group have indicated that RG-58 cable produces no impedance mismatch so you should start by doing one local loop and see how it works. Naturally I'm assuming that your current Ethernet isn't too long to add loops. -- John Cornelius (...!sdcsvax!piaget!jc)
flake@ncrcae.UUCP (Joe Flake) (12/12/86)
> >We're using garden variety RG58/U cable with crimpon BNC connectors, >ordinary N to BNC adaptors, and N barrel fittings. Nothing special. > BE CAREFUL! The cable you called out "RG58/U" is rated at 53 ohm impedence. It WILL cause noise problems. What you want is "RG58C/U" which is 50 ohm. We use thick to thin cabling with N to BNC adapters with good success. (We even used the "wrong" RG58 until some of our HW types started looking closely at the E'net interface!") Joe Flake NCR Corp., W. Columbia, SC ...decvax!mcnc!ncsu!ncrcae!flake
eriks@yetti.UUCP (Eriks Rugelis) (12/14/86)
i would like to thank those of you who have posted your about experiences with 'butting' thick and thin ether-cables. you have removed at least part of my fears in this regard. however, In article <136@piaget.UUCP> jc@piaget.UUCP (John Cornelius, System Manager) writes: >There should be little or no penalty for interrupting your thick cable and >running a loop of thin cable to the workstations. What may be most >appealing might be to leave the thick cable as a backbone and then have a >small number of fairly long local loops as in: > >=======| |==============| |============| |=== > | | | | | | > X X X X X X > | | | | | | > X X X X X X > | | | | | | > --X--- --X--- --X--- . . . >-- >John Cornelius >(...!sdcsvax!piaget!jc) so what about the connection spacing considerations that are laid out in the ethernet standard and that cause thick ether-cable to be 'marked in a contrasting colour' at certain specified intervals? my (probably limited) understanding of the physics of the situation leads me to believe that this is to avoid standing wave patterns in the cable (is this really off-base?) if you intend to interrupt your thick ether-cable to insert segments of RG58, do you scale the segments to n * ((185/500) * thick cable node spacing) ? do 'tails' from BNC T-connectors present in the 'loops' constructed in john's example have to be considered in the thin-segment length equation? i realize that a lot of people out there don't mind just plugging things together to see what they can get away with. i am one of those people, whenever i can find the time. however, debugging an ethernet after it has been put in place and users have come to depend upon it is not my idea of time well spent. in this instance i would rather be aware of the limits BEFORE i start pushing them. --- Voice: Eriks Rugelis Ma Bell: 416/736-5257 x.2688 Electronic: seismo!mnetor!yetti!eriks.UUCP or eriks@yulibra.BITNET Would like to be: Eriks.Rugelis@york.ac.can Quote: User-friendliness is one of the worst forms of brutality known to man. -- Voice: Eriks Rugelis Ma Bell: 416/736-5257 x.2688 Electronic: seismo!mnetor!yetti!eriks.UUCP or eriks@yulibra.BITNET Would like to be: Eriks.Rugelis@york.ac.can Quote: User-friendliness is one of the worst forms of brutality known to man.