Ethernet Network Taps

This approach would introduce errors at line rate would it?

snipped-for-privacy@gmail.com wrote in part:

I would expect so. It basically introduces a Tee [fork] into a transmission line that was carefully designed without one. It might work if the branch is kept very short. Or maybe not, since the termination will be wrong.

-- Robert

Not having looked at the previously mentioned site, yes, you want no tees, or at least minimum length of the side branch, and want no terminating resistor on the branch.

Normal transceivers (and repeater/switch ports) should have a terminating resistor of 100 ohms.

A three (or more) port switch with the resistor removed on two ports, and the two data pairs of a Cat 5 feed-through cable connected to those ports should work. One should probably also terminate the transmit pair for those ports.

I don't know how easy it is to get to the resistor on current 10/100 ethernet switches.

-- glen

Agreed.

And I'll bet this is built-into the MAU chip with it's isolation transformers. I doubt it'd be easy to scratch off the PCB on either the hub/switch or the snooping card.

The Tee arrangement may still work with extra 100 ohm terminators iff the cables are short.

Best would be a simple 100 _hub_, but those are now rare.

-- Robert

Robert Redelmeier wrote: (I wrote)

Resistors are not so easy to make on ICs, but they might do it by now.

If you really want to do that, you want the termination slightly higher. A small series resistor would probably do it, too, though that complicates keeping the signal balanced.

Find a 3C250 on eBay. They are fairly cheap, though shipping might be high. You probably won't find a more ruggedly built

I believe that was one of the first 100baseTX repeaters around. (I once bought one used, but I never opened it to see.)

-- glen