Thicknet/10base5 Test Segment: The Cable is In!

[Paul Koning]

> On Jun 26, 2018, at 1:19 PM, Chuck Guzis via cctalk <cctalk at classiccmp.org> wrote:
> 
> On 06/26/2018 10:04 AM, Grant Taylor via cctalk wrote:
> 
>> My assumption was that "tap" comes from the second form.  I always
>> thought there was a different name for the first form.  But I believe
>> they were less common, hence fall under the "tap" term which is more
>> popular.
> 
> My impression from the old days of this system was that the so-called
> "vampire taps" were superior in that they caused less of an impedance
> "bump" when attached, as compared to the cut-and-connector-only type.
> 
> Perhaps that's not true--can anyone verify this?

I haven't heard that, and it sounds questionable.  All taps are impedance bumps because they attach a stub to the cable.  The spec limits the length of that stub for this reason.  But N connectors are constant-impedance types, in RF service rated up into the GHz range, so the connectors should certainly not be an issue.

If a cut type "tap" were designed as a substantial size PCB with a connector at each end and wires (rather than transmission line) in between, that would certainly be bad, but that's just an example of the fact that you have to use transmission line design techniques when dealing with transmission lines.

A cut design could allow for shorter stubs than is mechanically feasible with a vampire tap, so if anything it would seem that the cut design has the potential of being better.  But it doesn't really matter; a correctly built vampire tap installed properly will give you a compliant Ethernet, and a thicknet segment can be quite large because the design was done so carefully.

	paul