Possible to have too much magnetics in Ethernet

Question

I have a question regarding magnetics topology on an ethernet 1000base t channel. I'm working on a project where I have a board (board 1) with MAC, PHY, and magnetics.

I have another board that mates to this (board 2) through a board-to-board connector. On "board 2" i'm taking the ethernet channels (4 differential pairs per channel) and routing them to RJ-45's with built-in magnetics. Then from the RJ-45's with built-in magnetics, the ethernet goes through standard cabling to an ethernet port on a PC (with built-in magnetics). So, if we follow the ethernet channel it goes

board 1 -> magnetics -> board 2 -> magnetics -> CABLE -> magnetics -> PC

Will 3 magnetics cause noise/oscillation or disturb ideal impedance matching?

According to "https://www.we-online.com/web/en/passive_components_custom_magnetics/blog_pbcm/blog_detail_electronics_in_action_85119.php"

A typical LAN has 2 magnetics per ethernet channel. I haven't found any resources on multiple (more than 2) magnetics circuits on a given ethernet channel.

Answer 1

The topology you have listed is a little difficult to understand. The proper way of doing this is to make sure you have a phy on each end of the magnetics.

This is how I would consider designing the system:

If board 2 is simply a pass through:

Board1 phy-> diff pair 100Ω and 100Ω connectors->board 2 -> magnetics -> cable

OR put the phy on board 2, it will increase the conductor count on the cable, but might be a better option as the phy's signals run at slower speeds.

Board1->phy on board2->magnetics ->cable

If you are doing anything with the signal on board 2, then that will probably violate the 100Ω impedance (you can't split a transmission line easily.)

Also a typical ethernet channel has one transformer per line and one choke.

Check out these questions: Connecting two Ethernet PHY without magnetics? Connect two embedded devices with USB host with Ethernet?