4
\$\begingroup\$

I'm working on a product design which includes an Ethernet jack (with integrated magnetics) that will need to be remote from the controller. The controller and jack are contained in different parts of the product that rotate relative to each other. Because of this I was thinking to use FFC cable. The total cable distance would be 10-15cm. I'm fine with only 10Mbps so I thought it should work OK.

I was planning to alternate signal with ground in the cable (as described here: I2C on FFC or IDE cable - interference?) but I'm a little concerned about the impedance of the FFC. I would normally use a calculator like this https://www.eeweb.com/toolbox/microstrip-impedance to determine trace widths, but they assume a ground plane and I wanted to use a standard FFC cable with no shielding. What would the optimal pitch and conductor width be to achieve 50Ω impedance, assuming the signals were G S G S G, etc.?

The controller would likely be something like the LAN9514, with integrated MAC and PHY. If the above approach is a major problem, would splitting up the MAC and PHY (placing the PHY by the jack) be a better solution?

\$\endgroup\$
  • 1
    \$\begingroup\$ Why not put the '9514 and jack on a board you can connect to the chassis and then use an internal USB cable to connect them? \$\endgroup\$ – Ignacio Vazquez-Abrams Dec 2 '16 at 23:16
1
\$\begingroup\$

A simple cheap, easy and elegant solution would be to put the jack on the board next to the phy and run an ethernet extension to the outside of the product enclosure. There are even extensions available with built in holes for fasteners for easy installation and manufacturing.

You might be able to get a double sided flat flex cable, but flat flex cables are not cheap unless you are buying them with quantity. And then you have to have another board at the output and control the impedance all the way from your board to the other mini PCB.

Same goes for USB, is easier to put the connector on the board and then run an extension cable then roll your own.

\$\endgroup\$
  • 1
    \$\begingroup\$ While there's a simplicity in that approach, it often isn't mechanically workable as the usual form of these connectors are fairly large and may not fit within the product. Also trying to bend the usual form of the cables too close to the connector may put side loads on the assembly which can prompt early failure. \$\endgroup\$ – Chris Stratton Nov 1 '16 at 18:06
  • 2
    \$\begingroup\$ Yes, this is the main issue, there is not enough space where the controller is to place a standard RJ45 jack. \$\endgroup\$ – AngeloQ Nov 1 '16 at 20:45
1
\$\begingroup\$

The characteristic impedance of common FFC is unlikely to be a good match for Ethernet signals. However, if 10Mbps is sufficient for your application, and this is not a commercial product which needs emissions certification, then I think it would work fine over 10-15cm of FFC as you suggest.

The threshold at which it "works" is dependent on the sum of all impedance mismatches between your PHY and the other end of the link. So if the cabling from your switch is already marginal, adding your FFC could push it over the edge to failure. Moving the PHY adjacent to the jack (with appropriate transmission line/impedance control between the MAC and the PHY) would be the best approach. But at 10Mbps there is a huge margin available, so as long as you don't have speaker wire or coathangers elsewhere in this link then using a short length of FFC might be considered good enough.

\$\endgroup\$

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.