I'm currently working on protecting 1000Base-T Ethernet lines from EMP using filter connectors such as described on page A-7 of this doc:


One manufacturer can provide a minimum capacitance of 30 pF from each connector pin (TX+,TX-,RX+,RX-) to the connector shell which is connected to chassis.

I'd just like to know if anyone sees any trouble here? I don't think they'll load down the signal since they're connected to chassis and not across the lines.

But I might be wrong. Thanks.

  • \$\begingroup\$ They won't affect the data lines. They are meant to capacities timely decouple the chassis to say earth. This is mainly when you go to plug or unplug the Ethernet cable , but f you discharge, the energy will capacitive my couple to earth instead of all over the place, just make sure you have an earth terminating scheme and wide flat low impedance traces to it. Good luck \$\endgroup\$ – cowboydan Sep 18 '15 at 0:52

An Cat. 5e UTP has 3 pF/m, or 300 pF / 100 m. Using this filter (like any other) you decrease the possible link budget, but this is the cost of protection. Why not? Or is it so hard for your application? Compare with 50-60 pF/m for S-STP.

Also, Ethernet has 100R at both sides in parallel, resulting in 50R equivalently. Both sides protection Tau = R*C = 50*2*30E-12=3E-9, i.e. 1/Tau is about 333.3(3) MHz, therefore it does not (significantly) limit the connection.

| improve this answer | |
  • \$\begingroup\$ pF / m is not a useful figure for a transmission line. CAT5 cable is 100 ohms and when terminated, looks like 100 Ohms. No capacitance visible. \$\endgroup\$ – tomnexus Oct 8 '15 at 19:38
  • \$\begingroup\$ @tomnexus An idealized transmission line is characterized by its impedance. To work properly, 1) transmitter output, 2) receiver input and 3) transmission line impedances must be all matched, yes. What did you mean speaking that "CAT5 cable is 100 ohms [diff] and when terminated, looks like 100 Ohms"? Does it look another when not terminated? Really? I think this termination is the property of transmitter and receiver, not the line --- but to match to the line. \$\endgroup\$ – asndre Oct 9 '15 at 5:08
  • \$\begingroup\$ @tomnexus Dealing with protection, you need to analyze to factors (at least): attenuation and distortion. You can use very complex analysis (as you wish) or use a simple one. \$\endgroup\$ – asndre Oct 9 '15 at 5:11
  • \$\begingroup\$ I'm trying to say that a (terminated) transmission line looks like a pure resistance to the transmitter. No capacitance visible, it's completely compensated by the line inductance. So it makes no sense to compare the lumped capacitance of the protection device to the distributed capacitance of the line. If the other end isn't terminated, then a line looks like an inductor or capacitor, depending on its electrical length. \$\endgroup\$ – tomnexus Oct 9 '15 at 5:17
  • \$\begingroup\$ @tomnexus Also, keep in mind please that the filter capacitance is related to the transmission line, very "centralized" and not distributed over the line and that with an Surge/FT/ESD spectra the transmission line is not so matched as with the Ethernet signal spectra. Therefore in the case of EMP you cannot assume the transmission system (transmitter, line, receiver) as a matched impedance transmission path. \$\endgroup\$ – asndre Oct 9 '15 at 5:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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