[Gb-ccb] CCB interface questions

[Martin Shepherd]

On Thu, 11 Mar 2004, Brian Mason wrote:
>...
> I see. So is it correct then that one overflowed ADC sample would set the
> entire integration to 2^32-1?

Correct. I can't think of any reasonable alternative. You can't just
drop overflowed samples from the parent integration, since this would
give that integration a different duration than the integrations in
the other phase-switch states, and thus compromize the cancellation
properties of phase-switching. Similarly, if you ignore the ADC
overflow, and simply add in whatever number it comes up with, then the
final integration will have an incorrect value. In other words,
whatever one does, the final integration will be corrupted, so it
makes sense to flag it with an obviously corrupt magic-value, rather
than present a subtly corrupted value that might go undetected.

Martin