Ethernet Routing between PHY and Magjack

Answered
0
0

Hi All

First off I’m really impressed by an open source project of this calibre – and the great tutorials!

I am designing a custom baseboard and unfortunately the PCB manufacturer I use has a standard 4 layer stackup with 0.308mm between the outside layers.

I am routing ethernet signals from the PHY (connector on the module) to a magjack. With a .308mm thickness, my traces need to be 15mil with 15mil spacing which gets me Zdiff = 98 Ohms and Z0=57 Ohms. This is within spec. When routing, the pairs and traces in the pairs need to be within .127mm of each other or so. Since the traces are so think, meandering and what not needs to be quite aggressive to affect the length which feels like it will be an issue?

Does anyone have any advice / tips with length and impedance matching on thick traces?

Thanks

  • You must to post comments
Best Answer
0
0

Hi Mitch,

yes, the distance between the waves needs to be at least three times more than differential pair spacing. Have a look on the design guide / datasheet of your PHY chip. There should be detailed information.

If you will use same topology (same layers, number of vias) for all signal within a group, the calculation of Altium is sufficient. But you can double checked it when you select all segments of the signal and use Reports -> Measure Selected Objects. The number will be smaller comparing to the length in the PCB bar because Altium sum also a length in the vias.

Yes, the length matching of thick differential pair may be sometimes challenging. :)

  • You must to post comments
0
0

Hello Mitch,

do you mean that the length matching will be a problem?

I guess the length matching within the differential pair do not really depends on the track width a pair gap. But the length matching of all differential pair may be a problem because you should need to leave quite much space for the waves and other differential pairs (clearance 3-5 times the pair spacing).

I think there is no special tip for length matching of thick differential pairs. But the position of the components and enough space are more crucial then with thinner pairs.

  • You must to post comments
0
0

Hi Martin

So the ‘length’ of a wave for length matching needs to be quite long (say 3 times the thickness of the track or the spacing?)? I have seen this before in a presentation but I’m not sure.

But yes the length matching is hard becuse I’m not sure how accurate Altium is with the calculation with thick tracks. Thin tracks are almost 1 dimensional so it’s easy!

  • You must to post comments
Showing 3 results
Your Answer

Please first to submit.