Increasing the Termination Resistor value for Ethernet

[FreshmanNewbie]

Suppose I am using a current mode Ethernet PHY which requires a termination resistor of 49.9 ohms to VDD on the Ethernet differential lines.

Can I increase the resistor value to 75ohms? Is it OK? Can someone tell me what impact it would have?

Request to provide an answer in simple terms

 

[KlausST]

Hi,

First: Why?
If you don´t have a good reason (and the knowledge what will happen), then simply don´t do it.

Impact:
On impedance mismatch you get echoes. It means the signal you send will be reflected. One "edge" sent will cause several additional edges on the wire/trace. Depending on the magnitude of the echoes the data may become currupt ... up to complete imposssibility of reliable data transfer.

50 Ohms transmitter --> 50 Ohm PCB traces --> 50 Ohms cables --> 50 Ohms termination.

Mind: This is characteristic impedance (cable, traces) and can´t be measured with an ohmmeter.

Klaus

 

[FreshmanNewbie]

Hi,

First: Why?
If you don´t have a good reason (and the knowledge what will happen), then simply don´t do it.

Impact:
On impedance mismatch you get echoes. It means the signal you send will be reflected. One "edge" sent will cause several additional edges on the wire/trace. Depending on the magnitude of the echoes the data may become currupt ... up to complete imposssibility of reliable data transfer.

50 Ohms transmitter --> 50 Ohm PCB traces --> 50 Ohms cables --> 50 Ohms termination.

Mind: This is characteristic impedance (cable, traces) and can´t be measured with an ohmmeter.

Klaus

Thank you for the answer. So, what is the one reason which the main reason on why we should not increase the termination resistor values?

 

[KlausST]

Hi,

So, what is the one reason which the main reason on why we should not increase the termination resistor values?

I don´t want to change it. I don´t have a meaningful reason.

You want to change it. Thus I guessed you have a reason.

Klaus

 

[dick_freebird]

If you go to 75 ohms you will see "staircasing" on the line
(not a clean transition, but multiple time-of-flight-and-return
"ratcheting" toward final value. This can make reduced
noise margin, bogus / bistable edge detection or sampling
at the receiver, etc. Long cable runs will take a long time
to complete a bit transition. Intersymbol interference is
likely to get messy too, on a corrupted-match line.

If you are after a standby-current reduction, you could
consider a capacitor-blocked 50 ohms, with a higher-Z
DC shunt to maintain resting voltage where you want it.
Seen space customers do this on some of my old RS-422
parts, way back when.

 

[FvM]

You would use 2x75 ohm termination if your cable has 150 ohm impedance. 150 ohm unshielded twisted pair has been used in early ethernet systems, all recent ethernet cables are shielded 100 ohm, thus using 2x50 ohm termination.

It's specified for 10, 100 and 1000 Mbit ethernet standards.

 

[crutschow]

Look up "transmission line".