2

I am trying to create a wire-tap for ethernet cables by modifying a classic phone wiretap which uses inductance. I haven't studied this in quite some time and need to come up with a solution for a course project.

I have tested the wire-tap on earphones... put the coil around one wire going to an earpiece on a set of earphones and the wires from the coil went into an input on my amp, I was able to hear the music (very clearly) through the stereo speakers - so I know the concept is sound.

What I want to do is use this on the TX-wires of an ethernet cable (one induction coil per wire). The two coil-output wires need to feed the orange wire going to the sniffer machine and the two coil-output wires from the white-orange cable need to feed the white-orange cable on the sniffer machine. This is the part I'm stuck at (well for the moment)... I have 4 wires to feed 2, I tried to ground one of the wires, but that didn't work... has anyone got any ideas... or be able to help point in the right direction?

JYelton
  • 34,119
  • 33
  • 145
  • 265
user26119
  • 21
  • 2
  • Ethernet uses [differential signaling] (http://en.wikipedia.org/wiki/Differential_signaling), which makes this a little more difficult to do than for a standard single-ended audio line. Even disregarding that, it's going to be more difficult as you are dealing with 100MHz vs the 20kHz of the earphone signal. – us2012 Jul 07 '13 at 21:55
  • Can you post a diagram of what you're actually planning to do? Also I don't want to put you off, but I'm not sure how viable solution for Ethernet this would be, due to the frequencies involved and signalling methods. Also what would you use to decode the data? – AndrejaKo Jul 07 '13 at 21:57
  • What I wanted to prove was that there could be a way to tap an ethernet wire without actually touching the copper. While I understand (more and more) how difficult this is going to be (and may now cost me an extra year in collage), I just want to prove the concept on 100M. What I'd like to do is induce the signal off both TX and RX pairs, possibly amplify them, pass it through a hub (as you would a degraded signal at ~100m) and then snoop the traffic from another PC connected to another port on the hub.I understand the diff-signals, but am unsure as to how the freq will cause an issue? – user26119 Jul 08 '13 at 10:50

2 Answers2

2

This is not a reasonable do-it-yourself project if you're at the level that you have to ask here. Go get a hub.

The reason this isn't as simple as tapping the magnetic field from headphone wires is that ethernet needs to stay carefully impedance-controlled, and the signals are a few orders of magnitude higher frequency. The first means you can't just magnetically tap into the ethernet signals without possibly messing them up in the process. The second means you have to know a lot about high speed circuitry to design something that could pick up the weak signals from the tap even if that did work. Explaining all the high frequency electronics you would need to know for that is well beyond what can reasonably done in a single answer here.

Again, go get a hub instead.

Olin Lathrop
  • 313,258
  • 36
  • 434
  • 925
1

I'm still a bit unclear on what you're doing (what is the tap connected to?), but I'd like to point out a simpler solution if you really don't want to do it "properly" with a hub: you can splice another ethernet cable onto the middle to create a cable with three ends.

Certainly this is bad for the signal integrity, but if the cables are short (few meters) this works OK. Caveats:

  • need to turn autonegotiation off on the "tapping" card
  • all this only works at 10 or 100Mbit
  • you can tap one direction at a time

Doing it inductively might be possible - Ethernet is inductively coupled at the ends, after all - but I doubt it would work with a tapping coil sized for audio frequencies. The approach to try there would be to take one of the tapped pair and connect the ends of its tapping coil to the recieve pair on the sniffing machine. Debugging such a system requires a suitably fast osciliscope.

pjc50
  • 46,725
  • 4
  • 65
  • 126
  • Thanks, I'll try that tonight. I was using 28SWG for the audio tap. I've got 40SWG to see if that works better for ethernet.The snoop machine is linux, with the interface in promiscuous mode. – user26119 Jul 08 '13 at 12:12
  • The "size" is not so much the diameter of the wire but the number of turns (and resulting inductance of the coil). I'm not an expert on coil construction and it's a tricky subject; have a look at http://drum.lib.umd.edu/bitstream/1903/11520/1/Bowen_umd_0117E_12074.pdf - the transformers it refers to are normally at the end of each piece of eithernet. – pjc50 Jul 08 '13 at 12:18