I'm using a N5171B to generate 50-MHz pulses, about 1us wide and 10us apart, and viewing a pulse on an oscilloscope. I've noticed that for an arbitrary pulse period (say 9 or 11us), the phase jumps around (for one pulse, the leading edge may be positive, and for another pulse, the leading edge may be negative). If I set the period to exactly 10us, then I get a constant phase. Is there a way to keep a constant phase for an arbitrary period? Or, if not, what are my possible choices of period to ensure that the phase is constant? Perhaps multiples of 10MHz?