1
\$\begingroup\$

enter image description here

I am wondering the possibility of crosstalk between the two ethernet lines and an analog signal.

The ethernet will operate at 100MHz. If I place positive and negative traces in parallel and place them sufficiently away(is 5 times the Width of the trace enough?) from the analog signal trace, will it prevent it from affecting analog signal majorly?

\$\endgroup\$
3
  • 1
    \$\begingroup\$ There is not enough information to answer. What is the impedance of the analog node? What is an acceptable noise voltage on the analog node? BTW 100Mbit ethernet does not switch at 100Mhz - it is closer to 31Mhz. \$\endgroup\$
    – PaulB
    Commented Oct 6, 2015 at 6:18
  • 1
    \$\begingroup\$ Ethernet is a broadband signal. A good rule of thumb for the bandwidth for any 8b/10b encoded signal in baseband mode is F(nominal)/5 to 2F(nominal); keep this in mind when considering interference from this type of signal. With more information on the nature of the analogue signal we may be able to help. \$\endgroup\$
    – Peter Smith
    Commented Oct 6, 2015 at 6:53
  • 1
    \$\begingroup\$ @Peter The Analog signal will be 10V output for measuring resistance ratio. It will have output impedance of less than 10 Ohms \$\endgroup\$
    – Steve
    Commented Oct 6, 2015 at 7:23

1 Answer 1

Reset to default
3
\$\begingroup\$

A signal with less than 10 ohms impedance will be susceptible to magnetic mode interference, and although the Ethernet pair will not radiate much common mode energy, there will still be some, and most of this energy is magnetic mode. See this previous question and answer for more details on common mode radiation, the causes and some solutions.

Apart from separating the analogue signal by some distance (which is an effective method of reducing interference), it may be necessary to add a guard track between the Ethernet pair and the analogue signal (which could probably be considered DC relative to the Ethernet pair). See this excellent article for guidance on guard traces.

That is just the starting point. Do you have any simulation tools available? Some permit proper simulation for both pre- and post-layout; if you have such tool(s) available, I strongly suggest using them.

HTH

\$\endgroup\$
3
  • \$\begingroup\$ your link to the article throws an error about a Access Key. Does it require some kind of login or something to be open? \$\endgroup\$
    – ricardomenzer
    Commented Oct 6, 2015 at 12:01
  • 1
    \$\begingroup\$ If you absolutely must have zero interference, go with a 4-layer or more PCB and route the ethernet (or analog) between two ground planes. This does come with it's own set of "gotchas" however, as the impedance and other parasitics change slightly for internal layers. (See Stripline for more details.) \$\endgroup\$
    – rdtsc
    Commented Oct 6, 2015 at 12:03
  • \$\begingroup\$ Nevermind, the link is working now. \$\endgroup\$
    – ricardomenzer
    Commented Oct 6, 2015 at 13:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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