0
\$\begingroup\$

I'm attempting to design a circuit that is capable of simulating an Ethernet disconnection event for a piece of test equipment. There doesn't appear to be any reference material that uses back-to-back magnetics, and I'm concerned about signaling voltages going negative (and therefore, out of range of what the mux can handle), so I'm using 49.9K pullups on each TR pair. I've also connected the center taps to +3V3, which I'm not sure is 100% correct. Can anybody provide me with some feedback, hints, and critique, here?

Schematic

\$\endgroup\$
8
  • \$\begingroup\$ Please add magjack type, preferably a link to data sheet. Also between which two devices the connection is? What is the test device? Is it gigabit or only 100Mbps connection? Why disconnection needs to be simulated? \$\endgroup\$
    – Justme
    Jul 1, 2022 at 18:03
  • 2
    \$\begingroup\$ In my view There is no need for magnetics. The structured cabling supports typically 3-4 passive connections in a signal chain, for 100BASE-TX, this is to allow punchdown blocks and patch panels to be 802.3 compliant (in most cases). Relays should be sufficient for a disconnect \$\endgroup\$
    – crasic
    Jul 1, 2022 at 18:06
  • \$\begingroup\$ Why not 8 throws worth of rf relays? \$\endgroup\$
    – Bryan
    Jul 2, 2022 at 0:03
  • \$\begingroup\$ @Justme It's gigabit. The DUT is another Ethernet device (sorry, cannot give specifics) and the upstream device is a run-of-the-mill network switch. \$\endgroup\$
    – foxtrot
    Jul 2, 2022 at 16:16
  • \$\begingroup\$ @Bryan I'm leaning toward relays; I was hoping for something non-mechanical – but I think that might be the easiest solution. The other option here is PHYs in a repeater (back-to-back) configuration – which is a bit more complex. \$\endgroup\$
    – foxtrot
    Jul 2, 2022 at 16:18

2 Answers 2

2
\$\begingroup\$

Just to make things clear: the following paragraphs only apply if you want to actually make something. You don’t need to. If the connection originates in a PC/Mac, you can programmatically disable the ethernet port and it’s equivalent to unplugging the cable. If the connection goes through a dumb switch, turn it off. If the connection goes through a managed switch, use the SNMP client to toggle the port enable state.

The simplest way to do it outside of the DUT is to use relays, believe it or not. Fast Ethernet works just fine through a QPST reed relay, or a couple of them side-by-side. For example, COTO 9011-05-11 would be a good choice.

The way you wanted to do it requires a repeater: MAGJACK - PHY - PHY - MAGJACK. The entire repeater can be powered down, and then it acts as-if the cable was disconnected. You could cobble it together from two PHY breakout boards, e.g. look for a LAN8720 Ethernet Breakout Board.

\$\endgroup\$
0
2
\$\begingroup\$

The Ethernet bus is supposed to have terminations at both ends of cable.

Your device adds two more so there is now four terminations.

It is unlikely to work at all.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.