I read this analysis skeptical of the satellite doppler interpretation. One thing I have not heard mentioned is jitter and drift in the frequency of the crystal used to create the pings.In cellphones, a cheap crystal is used which cannot hold a frequency at all. So each time it registers with a tower, it must first adjust for its own error to synchronize with the tower frequency. It is possible to spend a lot of money to make very stable oscillators for both the plane and the satellite. But no matter how much money you spend, it seems safer to give the plane the ability to adjust its frequency to match the satellite. This is especially true considering a slight drift in the frequency of the oscillator on the satellite could not be fixed. Unless the satellite were also self-tuning, to match a ground frequency in Perth.In any case, it makes sense that either the plane, or both the satellite and the plane, can tune their frequencies to synchronize with partners. It sounds like the Inmarsat system has been around a while. So it is likely that rather than trying to achieve perfect tuning, its protocol allows for a large amount of jitter and drift, rather than trying to get perfect synchronization. It therefore seems possible that the allowed jitter or drift between pings acceptable to the system, is around the same size as the frequency variation attributed to airplane speed.The round trip time would not be affected. Therefore, the distance of the plane to the satellite would be correct. But the shape of the arc, and therefore the location of the plane at that distance, would be wrong if the oscillator frequency has drifted. The search area would be much wider. Lucky oscillator errors could have made those pings zig-zagging around the South China Sea near the original flight path. Or the radio could have been in the water in a swift current, or in a fishing net, I don't know the minimum rate of movement away from the satellite.I am sure these smart guys know the frequency error of their own equipment. Or who knows, maybe they don't. To me, it seems like the plane will either 1) have an unknown frequency error if it does not tune its own frequency, or 2) will tune its frequency to match and cancel out the doppler shift. Maybe someone smarter than me can disagree.Estimating the error based on the frequency on the runway, or using the runway frequency as a baseline, could have problems. Frequency could drift significantly from small temperature and even humidity changes after the plane took off. Batteries or power sources could change voltage after flying for a while. The jet engines could have some resonant frequency that triggers increased jitter at certain speeds.In my experience, oscillator circuits with AC couplings are not random number generators, but like the surface of a lake which registers many small events. If the temperature of the capacitors is not controlled, it could perhaps change the apparent voltage at the crystal. If the structure of the circuit changes temperature during flight, it could deform slightly, changing the proximity of wires, and the accidental coupling between circuit components, feeding back into voltage and resistance, and resulting in a frequency change that could be equivalent to a few miles per hour in estimated airplane speed.One thing not mentioned in the analysis at the link, is possible error in the satellite frequency if both 1) it tunes its frequency to a ground station, and 2) this tuning was not designed to compensate for the decay in orbit where it is no longer stationary relative to the ground station. Again, this is too complicated for me to think about, so maybe somebody else can comment.Figuring in this error could be what is missing in the analysis at the link. But even if you know this error, you still cannot know the error on the plane. It might even be possible that the satellite resynchronized with Perth during the round trip back from the plane. And that there is no record of the size of the adjustment, other than what you can guess based on variations at Perth not accounted for by doppler shift from known orbit instability.When you figure in atmospheric interference, the error might be much bigger than the variations used to estimate plane speed, or it might be too small to matter. It is beyond my expertise.
Last edited by farmer
on May 9th, 2014, 10:00 pm, edited 1 time in total.