some confusion about the RF PCB trace radiation

I have some confusion about the RF radiation , some says when trace length of pcb reath to the 1/4 wave length of harmonics , it will radiate as a antenna , while xome says reath to the 1/20 wave length of harmonics for radiation . Which is right ?

Well, there is no actual defination for this. But What I can conclude (base on my experience) is if the distance of 2 parallel traces are 10 times the thickness of the PCB substrate, the crosstalk or coupling effect has been significantly minimized.

The radiation of a trace is dependent on the way the trace is loaded by the connected circuitry. A trace on a PCB is a transmission line. The same rules that apply to antennas and transmission lines apply to the PCB.

If the trace is poorly matched to the terminating loads, then there will be sufficient energy standing on the trace to radiate. If the trace is a resonant mutiple, or sub-multiple, of the operating frequency the trace will become an efficient antenna in the absence of a proper terminating load.

The short answer is - design the PCB trace with the proper impendance to match the source and terminating loads to avoid radiation from the traces. If your trace is a better match to the air or surrounding board structures than it is to the terminating impedances, you will transfer more energy to the air or surrounding structures than you do to the connected loads.

So how could i do to avoid pcb trace to radiate except to have a better matching ? when talked about the RF , it is always 50 OHM for almost RF componets , then how could i do it?