The solution frequency is the only frequency that is used for the adaptive analysis. I've never seen different results for a sweep when using a different solution frequency but I usually choose it to be near the middle of the sweep.
It looks like you did not decrease the frequency step size after the solution was found. To do this right click your frequency sweep >properties > and change the step size until you have a few thousand points. You cannot do more than 25,000 and it will warn you if you go above 100, just click OK and update the plot.
Side note: If you do frequency sweeps for parameterized results, for example one sweep for 100 different geometries, decreasing the step size will take a long time. I try to narrow the frequency band as much as I can so that I can get the same resolution for less computation time.