Simulated RF amplifier stability, but with a filter?

Hi All,

This has been really bugging me:  When I design and simulate an RF
amplifier (LNA) for a class project, it shows points of huge
instability (below the desired passband 'K' is *way* under 1 ).   But
when I add a bandpass filter to the input of the unstable simulated
amplifier, the instability is "gone" (the K is now way above 1).  My
question is this:  Is the instability really gone just because the
simulator cannot see the "real" S-parameters of the unstable amplifier
through the bandpass filter?  Is the filter just masking the
amplifier's instability when, in reality, the LNA is still highly
unstable, with or without the filter?  If so, is there a way to "see"
the "real" simulated K when a filter is placed at the input of an
amplifier?

Any help is most appreciated!

Thanks,

-Bill