Originally Posted by
Someone Somewhere
My understanding (from reading some of the raw data in previous FDR reports) is that the sampling is kind-of like old analogue video: there's always
something being sampled, and what is being sampled moves in a cycle. So the time between two samples of the same item is constant, but the time between two samples of different things relies knowing their position in the sampling cycle.
E.g. a very simple recorder might record the following:
0.0s Eng 1 cutoff
0.1s Eng 1 N1
0.2s Eng 1 N2
0.3s Eng 2 cutoff
0.4s Eng 2 N1
0.5s Eng 2 N2
0.6s left aileron position
0.7s left elevator position
0.8s rudder position
0.9s
stabiliser trim position 1.0s Eng 1 cutoff
1.1s Eng 1 N1
1.2s Eng 1 N2
1.3s Eng 2 cutoff
1.4s Eng 2 N1
1.5s Eng 2 N2
1.6s left aileron position
1.7s left elevator position
1.8s rudder position
1.9s
Flaps position
Note that:
- The time between successive measurements of the same thing is always the same
- In this case, most items are 1 second.
- Some items, in this case the trim and flaps, are sampled less frequently. The x.9 second slot gets used for one item on even seconds, and the other on odd seconds, so there's a 2-second sampling interval.
- The time between measurements of different items depends on their position in the cycle. Eng 2 cutoff is sampled 0.3 seconds after Eng 1 cutoff, but then it's 0.7 seconds before Eng 1 cutoff is next sampled.
- I suspect that the order of this sampling isn't strictly defined; it will depend on what the FDR/EAFR is selected to record and software version. I am not sure you can look at a previous report on the same aircraft type and assume everything is the same.
- If both switches were cut simultaneously at 0.2 seconds, the FDR would show Eng 2 being cut first at 0.3 seconds and Eng 1 at 1.0 seconds.
- If both switches were cut simultaneously at 0.7 seconds, the FDR would show Eng 1 being cut first at 1.0 seconds and Eng 2 at 1.3 seconds.
I don't know why the report says '01' seconds. I suspect this is just a typo.
Most interesting and I have a theory re the "01"
Without a decimal, the number implies a resolution of +/- One. That would be reasonable if it was reporting a movement, and not a "State". A condition....
What if the Cut was not generated by a physical movement, but by a Signal? Say a difference of 100 milliseconds? In theory, a differentiation in timing of a physical motion and a signal being computed in milliseconds? A command direct from software in the EEC to the Fuel MU without being detoured through the RDCs? (Or FADECs?) If an Overboost, wouldn't bits of seconds be crucial?
Thanks again.