> On Aug 26, 2016, at 11:00 AM, Ian McLaughlin <[email protected]> wrote:
>
>
>> On Aug 26, 2016, at 7:56 AM, Paul Koning <[email protected]> wrote:
>>
>>
>> That reminds me of an amusing error in a 1980s trade rag for DEC users ("The
>> DEC Professional"). It discussed how you could mix 10Base5 and 10Base2
>> hosts by coupling the two size coax cables through an N to BNC adapter.
>> That's fine so long as you use the 10Base2 limits. But instead of a barrel
>> adapter, the article had a photo showing a T connector at the transition,
>> with a "terminator" attached to it. So that network had three terminators:
>> one at each end and one in the middle. I suspect it didn't work, or at
>> least quite poorly.
>
> Are you sure about that? I’m by no means an RF expert, but, for example, the
> cable TV plant is a huge tree of cables and splitters. They advise that all
> open sockets should have a 75 ohm terminator attached to them. Even this
> whole mess with 75 ohm terminators everywhere, and the characteristic
> impedance of the network is still 75 ohm. Remember this is AC impedance, not
> DC resistance.
Yes, I'm sure of that.
Splitters are a different story. A splitter is a device that has one input,
terminated in the cable characteristic impedance, and multiple outputs each of
which are intended to be connected to a properly terminated line. So for that
reason, it is indeed correct that unused splitter outputs should be terminated
by the proper terminator.
But a T connector is not a splitter, which is why that analogy does not apply.
Ethernet (the baseband version at least) does not support splitters.
paul
