I think I finally got the rel nav filters under control (after some confusion what is what and updates what and what is displayed where...) So this is for fans of Shuttle avionics intricacies only. :lol:
***
We don't know precisely where the Shuttle is because the intertial navigation always drifts a little. So that mandates periodic state vector updates.
We don't know precisely where a rendezvous target is either for the same reason.
So the state vectors of Shuttle and target propagated by the navigation system have errors, and that leads to errors in relative quantities - range, proximity coordinates,...
On board sensors like star tracker or Ku antenna with radar ranging can measure relative quantities directly - all can do angular offsets, the radar can also do distance and rate of distance change. These pose constraints on the propagated state vectors, and if these constrains are used, they result in a filtered set of state vectors, from which a different set of relative coordinates can be derived.
But the system asks whether to incorporate a measurement by presenting the ratio - that's (filtered - propagated) value divided by some threshold - if the ratio is <1, the avionics believes it;s okay to proceed, if the ratio is > 1 the sensor might be malfunctioning and is better not incorporated - or the propagated value might be unusually bad of course.
Using the measurements and filters, the idea is to get closer to truth (which of course is easily obtained in the simulation...)
***
So here's an example of how the filters in REL NAV are operated to get a decent state vector (for comparison simulated truth is always shown in the lower left):
REL NAV is enabled and defaults to propagated state vectors. At this point the system thinks the range is 9189 ft, simulated truth would be about 9845 ft, so it misses by a good margin.
Radar ranging is now on and provides its own direct distance measurement of 9802 ft - this is well within the 80 ft uncertainty the radar has at this distance.
Now item 4 is used to show the filtered state vector and item 13 provides the angular data in addition to the range to the navigation filter. All residuals and ratios are okay, so items 17 and 20 are selected to automatically incorporate the constrains into the process. The system improves its estimate from 9189 to 9775 ft (which is doing quite a lot better...)
In the event, knowing truth we know the direct radar ranging measurement is in fact better - reality is that the propagated value is very low and the ranging measurement is low, whereas the filter thinks the propagated value might be too low and the measurement a bit too high - point being, it could be otherwise, we have no way of knowing in reality, so the filter does the best estimate given what we know.
Finally, transferring the filtered state vector to the propagated one zeros all residuals (by definition) and highlights the fact that zero residuals just mean that propagated and sensed state vector are identical, but they can both be wrong in the same direction.
Anyway, this is as good as it gets, so the range remains ~100 ft off. Luckily the station is in sight, and we can use COAS to do a size measurement once closer.
But never try to dock at night using radar ranging with the Shuttle...:lol:
