| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 372.1 | As the 3 year olds say...WHY ??? | PEACHS::SCHULTZ | I don't wanna pay SCHOOL TAX !! | Wed Oct 23 1991 23:25 | 10 |
|
I know this is a dumb question...But WHY...
A DEFZA to DEFZA via fiber does work, obviously nothing will take care
of any NOF's. Why will the STP/Thinwire not work the same way?
Is it because both transmit and recieve functions are happening on the
same piece of cable ???
MTS
|
| 372.2 | | KONING::KONING | Paul Koning, NI1D | Thu Oct 24 1991 12:52 | 24 |
| NOFs will be taken care of if the ring purger is enabled; ring purging
is NOT a function limited to concentrators! It may be that the current
firmware has it disabled by default because the hosts haven't gotten
around to providing management yet... but that's a different matter.
Anyway, the reason it doesn't work for the copper cables is because of
the cable detect circuit. Both coax and STP have a circuit that
detects the presence of the other node (and cable continuity in
between). For coax, this was done to avoid talking to yourself via the
reflection on an unterminated cable. For STP, it was done to avoid
talking to yourself across the loopback path of the ugly IBM token ring
plugs. Both are undesirable (the former especially) since it keeps
additional cable in the ring, and the path is not especially high
quality.
The cable detect circuit is essentially a voltage or current source
with load sensor at one end, and a sensor at the other end. It has to
be asymmetric (we tried hard to avoid it, but you just can't do it...).
Connect M to M or S to S and the cable detect circuit will claim that
there isn't a cable. If you were to disable that circuit, everything
else would work... but the price you pay is the loss of that function,
which was felt to be too high.
paul
|
| 372.3 | Any new STP cards in the future? | AUNTB::REED | John Reed @CBO - DTN: 367-6463 = DNIS | Fri Jan 24 1992 13:43 | 16 |
| How about M-M over the STP connectors ??
I understand that there is a difference in the concentrator portion,
and the workstation portion, to detect a completed circuit. Is there a
a card that could be used as a second-tier concentrator card for
customers that wish to have a high-density STP network?
Perhaps a top level DECconcentrator (with a Fiber DAS card, and two STP
cards), and several lower level concentrators with three STP cards, the top
card somehow being switched to emulate a workstation card. This would
allow a higher STP port density than needing a fiber card in each Conc.
I assume that this config is invalid, and cascaded concentrators must
all be cascaded using fiber. Leaving the bottom two slots for Copper
media attachments ??
|
| 372.4 | | KONING::KONING | Paul Koning, NI1D | Fri Jan 24 1992 16:12 | 6 |
| M-M is an illegal connection on ANY medium, that's a fixed rule of ANSI SMT.
In the case of STP, the electrical details of the circuit are a second
reason why that connection won't work, but even if you eliminate that
issue, M-M remains invalid.
paul
|
| 372.5 | Then, I'll try again !! :-) | AUNTB::REED | John Reed @CBO - DTN: 367-6463 = DNIS | Fri Jan 24 1992 18:39 | 8 |
| Sorry Paul,
Don't know what I was thinking... I was wondering more specifically if
we had perhaps an A/B implementation of STP. I assume from the
documents that the copper concentrator board is all M ports, and the
workstation cards are all S ports, and we don't have an A/B
implementation for copper, to allow us to cascade.
|
| 372.6 | | KONING::KONING | Paul Koning, NI1D | Tue Jan 28 1992 11:54 | 23 |
| The 5-company spec (which is what the current product uses) only define M and S
ports. I don't know if the ANSI standards effort will end up defining
other types as well, but there are a couple of reasons why we have the current
restriction:
1. Cable detect. The current circuit detects missing or broken cables.
This requires a different circuit at one end than the other, which is
one of the reasons why M-M is disallowed. The number of combinations
once you introduce A and B gets much larger. A-B is no problem, but
both A-M and B-M are supposed to be legal, so you can already see the
pattern.
2. Target environment. Copper cable, being strongly distance limited, is
targeted at the "last hop" in the wiring distribution tree. In other words,
in the equipment closet to desk hop. Those connections are also the
majority of total connections. The "higher up in the tree" connections
tend to be longer distance, and there are a lot fewer of them. The longer
distance says copper is inappropriate, and the small quantity says fiber
is affordable there (even if you have to do some rewiring -- since that's
often wire that goes easily into already-present ducts, and in any event
the number of runs needed is low).
paul
|