Antenna with long run using balun. Where should balun be placed?

[Alex]

We're designing a board with the nrf51822 QFA part. Due to design constraints, the chip will have to be about 15mm from where the antenna port will be. The routing to the antenna port will have a decent keepout and no planes directly underneath. It will be an M2M device, so there will be other interferers (enclosure, etc).

We have 2 options for the routing of the trace from the antenna pins to our port:

1. Route the differential pair as close to the port as possible and place the balun next to the port. Then route the single-ended trace a very short distance to the port.
2. The opposite. Run the diff pair immediately to the balun and then have the single-ended trace make the travel to the port.

My assumption is that 1 is the better option since it's a diff pair and should be more noise tolerant.

In either case, the single-ended trace will be routed with a 50-ohm impedance.

Thanks, Alex

[Asbjørn]

You should go for option 2and keep the differential tracks as short as possible and the balun as close to the chip as possible. Then from the 50 Ohm output you do a 50 Ohm trace to where you need it. We use AppCAD to calculate the dimensions on the 50 Ohm trace.

The problem with running the differential lines is that every millimeter of trace at 2.4 GHz on FR4 will be just about 1 nH of inductance and also introduce stray capacitance into the system. When the inductors used are in the low nH range themselves, the additional trace inductance become a big problem. So keep the balun as close to the chip as possible and the differential routing as short as possible.