However, bounces can be pretty short, hence you can miss 'em if you don't have ~120THz processor, which I certainly don't have, and I'd bet Suresh doesn't either.

The integrator relies on the fact that there's no current flow unless the switch IS closed. While the reed is bouncing, the charge rate of the integrator will be less than when it is in a stabile closed state. Its operation is also benefitting from the fact that, while the initial voltage, prior to the first contact "make," the voltage is known, hence it contains information about when the first make occurred. When the reed bounces "open," it simply doesn't conduct. The result, ideally, would be one charge rate while it's bouncing, bracketed between a constant zero before the first bounce and a previously known rate that occurs only after the reed has stopped bouncing. That way, the same loop can extract not only the "activation" time, i.e. the time from when the coil was energized to the time the reed stops bouncing, but the actual bounce time without the travel from open to closed state.

A lot hinges on selection of the proper integrator time constant.

RE