The signal is 20 mVpp, so that is the full ramp in the image. I just posted more in response to Bob Koller.
My calculation of the frequencies required is not correct. Not sure what I got wrong. With the settings I gave I get a flat trace across the screen which slowly rises and then starts over at the bottom. I'm guessing I calculated the frequencies such that each sweep was a constant offset from the trigger.
With a 100 kHz ramp I need a trigger around 99.989 kHz to display the full ramp for a single cycle. That's a difference in period of ~1.1 ns. Or with a 100 kHz trigger, I need a 100.0108 kHz ramp.
I've never dealt with such extreme aliasing before and normally I am considering the frequency domain effects. So this will take some thought.