Connect 2 Ethernet PHYs with a transformer with different CT voltages

[dyunasing]

hi, all,

I am connecting 2 Ethernet devices (one of them is a modem, the other is my DM9000B ethernet PHY). I did try the capacitor coupling method, but it did not work. It may be due to the auto-negotiation feature enabled by the devices. (**broken link removed**)
In the end, i have to use a transformer to isolate them. However, this method gave me around 5% failure rate where the MCU that connects DM9000B was not able to ping to the modem.
The DM9000B has a CT voltage 1.8V and 3.3V for the modem. The 50ohm pull up resistors are already built into the modem and i have no access to them.
I have a gut feeling of these different CT voltages causing this problem. Could anyone help to shed some light on this?
thanks.

Dyu

 

[KlausST]

Hi,

Don't bother about the center tap voltages.
On the receiver side it is just used to bias the inputs, so that the voltages keep within the common mode voltage of the receiver comparator.
On the transmitter side it is useless somehow (for totem pole outputs). Maybe the drivers use open collector output stages..

Terminating resistors should be on the receiver side (with the transmitters having internal series resistors). But on the bottom data line the terminating resistor is at the transmitter side.

Ethernet typically uses two transformer pairs. One pair at each device. But here only one pair should be sufficient, as long as the transfer ratio is correct.

Use Ethernet cable, shielded, twisted pair with defined wave impedance. Take care that the shielding generates no ground loop.

Klaus

 

[dyunasing]

Hi, Klaus,

Thanks for the reply. Am I right to say that you did not see a problem with the way i connected the 2 PHYs?
Those circuits are populated on the PCB.

 

[KlausST]

Hi,

yes, i see no problems.
But i didn´t read the datasheets either. If you are not sure, then comply to the datasheets.

Klaus