Background
My project is split over two PCBs for convenience.
The first PCB adapts an input signal and converts it to a low-voltage differential signal. There are 4 channels to be measured, so 4 differential pairs. Signals are 50 kHz, about 1Vrms. Seemed best to use a regular ethernet cable to carry over those signals to the other PCB (the data acquisition board). The cable distance is 1.5—2 metre.
The RJ-45 receptacles I selected have metal outer casing, exposed as through-hole pins. So if I will, I can make the whole ethernet cable's shield be electrically connected to a copper pour on both PCBs, which surrounds the area around the connector on the PCB. I can also elect to connect this copper pour to analog ground.
Options
So I have a multitude of choices:
- Don't connect the cable's shielding to the RJ-45 ends of the cable (or use plastic ends). This way the cable shield will be floating with respect to the signals it carries
- Use metal ends and connect them to the cable's shielding, but do not solder the PCB's RJ-45 receptacle to the copper pour. This way the cable's shielding will be connected to the receptacles, but still electrically floating.
- Solder to the copper pour (still electrically floating).
- Connect the copper pours to the PCBs' analog ground - the shielding becomes ground potential.
The PCBs already use the same ground, as they are powered from a common DC source. Option 4 creates another ground path which I'm slightly worried about.
Question
I'm wondering which option is the best from signal integrity standpoint? I want to measure the differential voltages with a 12-bit ADCs, so signal interference and noise could be an issue, hence the question.