FEDEVEL
Platform forum

kabaleeswaran , 03-14-2024, 04:19 PM

As i already discussed regarding differential pair impedance fedevel forum,
Differential impedance to be maintained with the help of ground plane .but still i need to confirm few point in Magnetic Ethernet differential pair routing,
How differential impedance calculated with out ground plane in Magnetic ethernet section. please check the below link.

https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/SupportingCollateral/LAN8742_QFN_Rev_A_Routing_Checklist.pdf

QDrives , 03-14-2024, 04:53 PM

I found this comment interesting:
You can use a trace impedance calculator for stripline and just make the H (height above the ground plane) very large. But make sure you get the spacing between the diff pairs and their thickness (copper weight) correct.
But it's going to high, probably over 100 ohms. Note that starts to look like a twisted pair configuration, that has a Zo of approximately 120 ohms.
As to adding a ground plane, it should be connected to signal/power return. A picture of your connector and pin assignments would help.


As was found here: https://electronics.stackexchange.com/questions/670006/differential-pairs-without-ground-plane

kabaleeswaran , 07-24-2024, 10:56 AM

Hi Qdrives,

kabaleeswaran , 07-24-2024, 11:01 AM

unfortunately I miss this loop. you mention Strip line but from this image itself, Differential pair routed on microstrip line. can you please check the same link from my query and reply. without ground reference how width/space defined..is that document is correct to refer ?

QDrives , 07-24-2024, 07:46 PM

"...*stripline*..." was a copy of a comment from the link I gave.

As per your reference, it does not state what width as spacing should be used.
It just states this: "*The traces connecting the transmit outputs from the magnetics to pins 1 & 2 on the RJ45 connector must be run as differential pairs. Again, the differential impedance should be 100 ohms.*"
Now the question is: how do you get 100 ohm without reference plane.
For the link I gave, it states to set H as very large. When I do that, I get extreme values (like > 9mm wide traces).

The picture in your reference shows about the same width as the ones with a return plane.
My advice if you remove the return plane: keep the traces as short a possible and keep the same width as calcutated for the other traces with a return plane.