Investigate why loading ephemeris sometimes hurts TTF

[PaulFertser]

Daniel wrote: "I've tried this numerous times with the right time (off by about/less than a second) and the ephemeris that the ublox just received a couple minutes before and the result was pretty much the same. The chip would get a fix after ~5 seconds which is way off and then it will take about 7 minutes until the next fix."

I can confirm that sometimes i see exactly this behavior. But i think i saw a proper fix several times as well (or was it a dream?). Moreover i confirm that old ephemeris even after loading into the chip don't hurt performance as they are discarded by the chip.

BTW, the drift of system clock is really huge (almost 2 sec / hour) but it shouldn't be a problem for a 4-hour validity interval (but the time should correspond to the GPS time at the beginning of it).

I've read through  http://www.u-blox.com/customersupport/gps.g5/u-blox5_Protocol_Specifications(GPS.G5-X-07036).pdf but didn't find any clues why it doesn't work well for us. The only difference i found so far is that they suggest the following sequence: AID-INI, AID-EPH, AID-HUI, AID-ALM (p.29). I tried it but it doesn't seem to make any difference.

Maybe we can ask other folks who have tried this before, what do they think about this weird behaviour? I've seen at least one ruby implementation and i think there exists something that ASU uses.

[jluebbe]

u-center logfile of strange startup

[PaulFertser]

Wonka on IRC said that using Assist Online data with scripts from the wiki he never experienced weird behaviour and that sometimes TTF was below 5s.

[mickey]

Making this a release blocker for ms6.

[jluebbe]

The data structure definition was broken, this is now fixed and i've reenabled ephemeris upload. We need to test if this fixes all problems.

[daniel]

This definitely helps here. I collected ephemeris of ~7-8 SVs and now reenabling GPS has a TTFF of about 16 seconds. We can definitely close this ticket, though more optimizations in order to improve TTFF might be possible.