Distinguish first between "ripple" as the sawtooth riding
on switched-mode power supply output, and other LDO
output behaviors (like small or large signal oscillation,
overcurrent protection "motorboating", etc.). A LDO ought
to reduce ripple from what's coming in, but many have
such poor HF PSRR that they don't do it very well. And
it's possible to pick the right (wrong) compensation such
that some frequency bends could see no attenuation or
even gain. But in any case if it's real inbound ripple the
frequency ought to be a dead match, while loop or
protection-function "activity" will be something else than
what rides on the source supply.
A couple of measurements ought to narrow down the
source and suggest "solution avenues".