Preamble: I have read this thread on single gravity well binary system eclipses. And this one on binary system views. And this one on a gas giant's moon's solar eclipse. And (now) this one on seasons in a binary star system.
I am co-writer of a historical fantasy project, and one of our elements is slightly science-based: the story world involves a planet in a binary star system. Specifically, that our planet orbits one star (very similar to that of Earth) which in turn orbits another star, much larger but much more distant.
One of our narrative events is an eclipse of the far star (A) by the near star (B), as shown below in a highly technical illustration which I just made in five minutes using MS Paint:
Needless to say, Not To Scale. Probably.
To an extent, detailed mechanics of how one sun eclipses the other are not vital for my purposes (although I'd be interested in learning from anyone with useful knowledge on the subject). What I'm principally interested in knowing is this:
How long might such an eclipse last, from the point of view of a planetary observer?
My intuition is that, if all three bodies were perfectly aligned on a plane, the eclipse period would only be determined by the planet's orbital period around B. As the planet orbits, the more distant A would appear to slide out from behind B and gradually cycle around to the "night" sky, eventually returning to the "day" sky to be eclipsed again as the planet completes its year.
In which case, I guess the duration might be a matter of weeks, maybe a month, depending on exactly how big A and B appear in the sky (if all this were happening on earth). So, first off, does that sound about right?
From this, what I would also like to know is the following:
1) Can the frequency of eclipses be made plausibly less regular than simply annual?
I'm thinking star-to-star or star-to-planet offset planes, or a wobble in the orbit of one star or the other, perhaps. Something that would make an eclipse predictable but less immediately obvious. And if it resulted in weird irregularity of where the two stars appeared in the sky, that would be Just Fine. The weirder, the better.
2) Is there a way that the duration of the eclipse could be made longer?
For example, would different directions of the two orbits have an impact on how it is perceived, or something like this.
In an ideal world, we'd be able to justify claiming something like the eclipse occurs once every twenty (or one hundred) years, and lasts for a year (or five) whenever it does. A ratio of 100/5 would be great, or a longer waiting period even, but I'd take 20/1! However, I'd prefer this to have some degree of rigour to it, so if there's no plausible way to extend it knowing that would be just as valuable.
Any assistance offered will be gratefully received, along with a salute in our project when the event in question takes centre stage.