2
\$\begingroup\$

I'm trying to understand this schematic: RJ45 ethernet with magentics

The ethernet lines end in a device with the input already 100 ohm terminated. The magnetics have a ratio 1CT:1 so the input of the rj45 Connector should have termination resistances of 50 ohm on each line like the ones in figure, that should be not mounted. Moreover there are this 22 ohm resistors that I don't understand. I read that put a 22 resistor is a rule of thumb for high speed signal termination but I don't understand their utility in this case.

\$\endgroup\$
3
  • \$\begingroup\$ Where did you get the schematic? I havent seen these on any ethernet schematics before. \$\endgroup\$ Commented Jun 15, 2017 at 10:34
  • \$\begingroup\$ It's a schematic that it was given to me and now I'm dealing with this schematic design that is already partially done but I have never implemented a ethernet interface and so I'm trying to understand it. \$\endgroup\$ Commented Jun 15, 2017 at 11:41
  • \$\begingroup\$ I haven't either seen any resistors like them before for the network connections of a PHY. They are probably not necessary. But you could leave them there anyways, if you are afraid they are needed. When the board comes, measure your signals and if something seems strange you can always replace them with 0 Ohm. \$\endgroup\$
    – nickagian
    Commented Jun 15, 2017 at 11:51

2 Answers 2

3
\$\begingroup\$

According to my experience, I also agree, those resistors there are not common at all. I haven't seen anything like this. I would expect such series resistors on the MAC interface signals, in order to reduce their energy and rise time, for EMC reasons.

What's more, Microchip also don't have these resistors in the evaluation board for the specific PHY that is used.

I can only perhaps guess, also according to the comment on the schematic (ETH_couples: 100 Ohm differential impedance matching, that whoever did the design, also made the PCB traces have on purpose an impedance not equal to \$50\Omega\$ or knew it would be the case. And with these resistors he made sure that there is a correct termination to the real impedance value of the traces. In a similar sense that also Andy aka says: the internal output impedance plus the external series resistor is equal to the trace impedance.

That is just an assumption, but I cannot think of anything better.

\$\endgroup\$
1
\$\begingroup\$

The lines that come in from the left and referred to as "ETH_TX" won't have zero ohm output impedance so, if they have an output impedance of nominally 28 ohms, that impedance plus the 22 ohm makes 50 ohms.

For the ETH_RX lines there may be a similar story or, if you had the full schematic that showed the chip interface, this might become clearer for both RX and TX lines.

\$\endgroup\$
3
  • \$\begingroup\$ The chip on the left side is the KSZ8051 ww1.microchip.com/downloads/en/DeviceDoc/00002310A.pdf and the datasheet says that the lines have onchip termination \$\endgroup\$ Commented Jun 15, 2017 at 11:05
  • \$\begingroup\$ The DS doesn't appear to state what the internal Z of txp or txm is. What page did you see it? \$\endgroup\$
    – Andy aka
    Commented Jun 15, 2017 at 12:27
  • 1
    \$\begingroup\$ It doesn't say the value but it says for example in page 4 that they use a onchip termination resistors for the differential pairs \$\endgroup\$ Commented Jun 15, 2017 at 13:06

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.