1

I would like to make a circuit that allows to detect activity on a RJ45 cable (on which Ethernet would be used). When the cable is only connected to a single computer, link test pusles are emitted. These signals are about 1 us long and between 0,5 mV and 3 V. Otherwise, data is Manchester-encoded, with a frequency of 100 MHz. I do not know the amplitude. I would like to detect, with a diode, activity on the RJ45 cable. As the signal is not continuous, a simple LED would not work. I also would like not to damage the signal. Do you have an idea on how it would be possible to do it?

pierre123
  • 49
  • 3
  • 7
  • 1
    Isn't that what the LED on the back of your computer is for... – Tom Carpenter May 10 '17 at 08:35
  • Are you trying to detect efforts to establish an actual Ethernet connection or just some other custom protocol being used on an Ethernet-style cable? – Finbarr May 10 '17 at 08:40
  • maybe a good hall-effect sensor can detect the differentials. – dandavis May 10 '17 at 08:44
  • Tom Carpentier : Yes, but I would like to do it remotely. – pierre123 May 10 '17 at 08:49
  • Finbarr : I am interresting in all solutions : detecting efforts to establish a connection, a real connection or both – pierre123 May 10 '17 at 08:50
  • 2
    Well, I guess it's possible but it won't be easy. The cable pairs are electrically isolated from the equipment by transformers at both ends, so you'd need to make sure you don't break that. You'll also have to make sure you don't put any significant load on the pairs or you'll disrupt its proper functionality. – Finbarr May 10 '17 at 09:21
  • Was sort of imagining putting a coil around it and looking for induced voltage... Don't know if that's a daft idea, whether it'd wreck the LAN signalling? – TonyM May 10 '17 at 09:43
  • Finbarr: what do you mean by "break that" ? (sorry, I am not an eletronics expert) – pierre123 May 10 '17 at 09:52
  • All : Thank you for your suggestions, they are all considered and appreciated – pierre123 May 10 '17 at 09:52
  • I mean by connecting anything to the pairs that results in them becoming connecting them to ground through a DC path. The pairs are twisted so they don't generate a great deal of magnetic field to detect, either. – Finbarr May 10 '17 at 11:04
  • TonyM, a coil around wouldn't hurt the signalling, but all it would detect is common-mode current (noise), which is dependent on the devices involved and can vary greatly. (Or, if it's PoE and the device is sourcing/sinking transients along an alternative return path, the common-mode current will be all over the place.) It might be possible to detect Ethernet harmonics with a sensitive probe, but it's not easy. – uint128_t May 10 '17 at 15:20

2 Answers2

2

As suggested by Tom Carpenter, the cheapest and easiest solution in Ghetto Engineering Mode will be to insert an Ethernet switch and monitor the LED.

bobflux
  • 70,433
  • 3
  • 83
  • 203
0

In the answer for this question one methodology to observe an activity (in the form of EM emission) on the cable is referenced and its results are depicted. From there:

Evaluated by measuring CM current on cable

RF current probe place on ... cable

You do not need a spectrum analyzer but an appropriate filter/amplifier+integrator. Cannot provide you with schematics but think the method may be useful for you.

Community
  • 1
asndre
  • 1,597
  • 10
  • 17