4
\$\begingroup\$

It seems that the voltage over ethernet ranges from 2 to 3 volts. How much current typically runs through the wires? I tried Googling with no success.

I ask this because I'm looking for components that are able to switch on and off an ethernet connection.

\$\endgroup\$
  • 3
    \$\begingroup\$ I assume you're not considering a power-over-Ethernet type of connection, which is typically ~48V and ~350mA. en.wikipedia.org/wiki/Power_over_Ethernet \$\endgroup\$ – HikeOnPast Jul 29 '12 at 0:36
  • \$\begingroup\$ Just to summarize, excluding power over ethernet, your primary challenge would not be the current-handling capability, but instead the maintenance of characteristic impedance through the switching path. \$\endgroup\$ – Chris Stratton Jul 29 '12 at 16:30
5
\$\begingroup\$

Physically switching Ethernet is hard to do, for the following reasons:

  • Ethernet is high frequency. In this area, you have to watch out for reflections of the electrical wave, which happens when the "characteristic impedance" of the transmission path changes. Ethernet cables have a char. imp. of 100 Ohms. You would have to design your switch so it has a characteristic impedance of 100 Ohm, but a specific (DC) impedance of 0 Ohm, like the cables.

  • The whole point of structured cabling is to remove physical connections in the cabling between the endpoints for this very reason. You connect each unit to a switch with a separate cable, so there are no disruptions in the characteristic impedance.

  • Ethernet is not meant to be physically switched. A network switch device does not switch the cabling, it does receive, manage and forward the packages.

About voltages and currents: Ethernet voltages mostly use -1 to +1 Volts. Currents would be in the range of a few 10 milli-Ampere, depending on frequency and length.

| improve this answer | |
\$\endgroup\$
1
\$\begingroup\$

Line is 50 ohm impedance - with driver IC's driving a 100 ohm impedance line balanced around ground with a transformer coupling.
Here is a typical driver IC data sheet TLK110.
They note a drive r power consumption of 250 to 275 mW in operation so actual circuit power will be below that.
At say 3V RMS equivalent into 50 Ohm power would be V^2/R = 180 mW so that's in the order of right.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ So current is 60mA RMS. \$\endgroup\$ – starblue Jul 31 '12 at 7:29
1
\$\begingroup\$

Are you sure you are asking the right question? There are lots of Ethernet standards, I assume your are talking about 100BASE-TX?

You can not directly put a DC voltage on an Ethernet connection using standard hardware (beyond other problems it is physically impossible because it is transformer-coupled on both sides). You can transmit a message (frame), which is a one-time thing. Or you can transmit repeatedly, but that still won't get you a DC signal. And if you transmit nothing, that will not guarantee that there is no signal on the Ethernet connection!

In short, you won't be able to do much with your Ethernet connection without an intelligence (microcontroller) on the oher side.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ I don't understand how this addresses the question. \$\endgroup\$ – Rocketmagnet Jul 29 '12 at 13:24
1
\$\begingroup\$

I know that on the market exists at least a switch that phisically connect/disconnect ethernet connections.

I've seen it once, it seemed a classic 12 port 10/100 switch, but with a special feature: the first two port are phisically connected when the switch is powered off, AND connected thru the switch logic when the switch is powered on.

I was rather curious and I've opened it, and there were two standard Finder relay whose NC contacts connected the rx and tx pairs between the first two ports.

Finder 3022

| improve this answer | |
\$\endgroup\$
  • 2
    \$\begingroup\$ You can often get away with interconnect that is far outside of spec - until you don't. From a product design perspective, the interesting question would be if they said "well, it still works with $(uncommon_cable_length), ship it!" or if they used test equipment such as an RF (not LAN) network analyzer to decide that the forward loss and reflection were within specification limits. \$\endgroup\$ – Chris Stratton Jul 29 '12 at 16:25
  • \$\begingroup\$ It's worth considering that telephone-style RJ45 punchdown connectors don't maintain characteristic impedance either, and work at least for the slower ethernet speeds (though there may be better designs for gigabit). At higher cost, there are of course carefully designed relays sold for actual RF switching applications, which might be found with the needed characteristic impedance, or used with matching transformers on either side. \$\endgroup\$ – Chris Stratton Jul 29 '12 at 16:31
  • \$\begingroup\$ @ChrisStratton I don't know how it's tested in the factory for the connection specifications, but I can tell you that that device was used in a place full of RF noise, and (said by the device user) never had a problem. Anyway I agree with all your comments. I've opened it because I had same doubts you've written so well... :-) \$\endgroup\$ – Axeman Jul 30 '12 at 12:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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