I need to assemble an accurately length-matched set of short coax cables. My target is to get a set of 1ns-long cables matched to 10ps or better.
I measured the cables using an HP8753D network analyzer to plot the phase of the S11 parameter of each cable in turn, the far end remaining open. I've adjusted the electrical length compensation to make the phase vs. frequency slope nominally zero. (Linear frequency scale, of course.) That gives me a family of plots like these: