1 PHY to 2 Ethernet magnetics

[jelydonut]

Hi,

I am trying to work out the best way to connect 2 sets of magnetics to the TX/RX of a 1 port PHY. I know it is unusual to do this. Basically it will look like below.

[PHY TX]
|
|--------------[10/100 Magnetics]----[RJ45]
|
--------------[10/100 Magnetics]----[RJ45]

(Note: it should "Y" out to the two magnetics from the single port of the phy, but the forum removes leading spaces, so its hard to draw)

I am having trouble working out what needs to be done to keep impedance, etc, from being a issue and losing signal quality or distance.

Thanks!

jelydonut

 

[FvM]

What should be the purpose of the Y-connection? Connecting two hosts at the same time?

This will hardly work because each ethernet interface is sending a permanent "idle" signal. The signals from two connected hosts will superimpose and probably none be received properly. Unlike with old coax "thinwire", there's no collision detection on twisted pair ethernet medium.

Similarly the Tx signal will be considerably attenuated by the Y-splitter and might be not received at all.

 

[jelydonut]

No, it won't be used for 2 hosts at the same time. It will only be 1 at a time. The key is that the ratio of each set of magnetics will be different, so the output impedance on the line side is different depending on the port. I was worried about a possible issue from Y'ing the differential pairs to each magnetics, but there should be a way around it.

Thanks

 

[FvM]

it won't be used for 2 hosts at the same time

So what's connected to the unused port?

 

[jelydonut]

Nothing. Only 1 port will ever be used at a time.

 

[FvM]

If the unconnected line length (Y upto RJ45) is short, there should be no problem. Parallel connected transformer inductance may distort the waveform a bit, but not seriously. What's the line impedance difference, 100 versus 150 ohms? The ethernet will most likely work with this impedance mismatch as well.

 

[jelydonut]

I figured they would have to be as close to the PHY as possible. One port will be the standard 100 ohm (1:1) magnetics, the other will be a 75 ohm (1.15:1). I still need the standard 300', so was worried that 2 "inductors" from each magnetic will be in parallel and possibly dampen the signaling. I may just have to proto it to see how it operates.