PJ2,
as always a pleasure to read ...
(Maybe this really belongs in another thread ...)
We measure 'g' at eight times a second, most other parameters at four, two or once per second and those which do not change rapidly over time, once every two or four seconds.
I've been wondering about the choice of sampling rates for different parameters. The Sungshan-Report showed that actual brake pressure was only recorded once every 4 seconds, although during anti-skid-operation it changes rapidly.
glob99 is of course right in saying that if the highest frequency that may be contained in the signal is low enough (half the sampling rate), data can theoretically be reconstructed completely, practically limited by the losses in value-discretisation on a digital recording and other technical and physical phenomena.
For values determined solely by the output signals of a computer, i. e. fuel flow, the position of control surfaces, etc, this frequency is determined by the software design and the processing speed. It can in theory be determined accurately.
Human inputs are a bit different, but given a certain inertia of limbs and controls, one can put reasonable upper limits to that as well.
Outside physical forces on the aircraft, as measured by the acceleration, may contain very high frequencies, especially if a hard object is hit, there is severe turbulence, or simply by a bumpy runway. In this case it is vital that these be sampled at a higher rates, e. g. to determine the exact location of a piece of debris on the runway that caused damage.
The simple on/off switching action (mathematically accurately described by the
Heaviside step function) can never be sampled completely accurate, since it contains frequencies up to infinite. A choice of needed accuracy has to be made.
I'm off to SAFECOMP in Nuremberg for the week now,
Bernd