Whats the typical observing efficiency in spectroscopic observing modes?
About 25%. A peculiar software feature requires us to wait an integral number of frames for the chop motion to settle. Thus, if your frame time is 0.5 s you must wait 0.5 s after the chop, (tossing this frame) then take a single 0.5 s frame, cutting effficiency. The alternate which may be acceptable sometimes is to wait zero frames after the chop and take the hit in the psf in return for higher efficiency.
