– Copied from a customer email –
We’ve purchased a number of GigaStax Rugged Ethernet switches and we’re developing a PCB to connect them to.
Table 3 of the Datasheet shows that there are a number of pins named GNDREF on the connector. There is no description about what these pins are or what they need to be connected to.
I assume GND but would like to be sure as there are also a number of additional GND pins on the connector.
Table 3 states that GNDREF is the ground reference for ethernet signals.
This is not the same as GND, which is the negative supply input terminal for the board.
The difference is due to the fact that ethernet signals are isolated by way of transformers, meaning there are two grounds to deal with.
GNDREF is typically connected to the chassis or shield of your device, whereas GND is the negative input supply terminal.
In most cases you would not want to connect GND to GNDREF, because doing so will mean that that the ethernet signals are no longer fully isolated from the ethernet circuitry on board side of the transformers. In some applications this is desired and that all depends how you are grounding your system.
If in doubt, it is better to not connect GND and GNDREF, you can add a small resistor and DNI it so you can later connect them if you find you need to.
The chassis is very often connected to the Negative input supply terminal is it not? So does it not amount to connecting GND and GNDREF ?
Yes, sometimes the shield and ground are connected, but not always and it really depends on the overall grounding scheme.
The shield connects to the sheath around the cable and will go as far as the next connected device. At that connected device, if it also connects the shield to it’s internal negative supply rail, you could have a rather long connection between two grounds resulting in a ground loop.
We find it’s better to keep shield isolated from ground to prevent this.