search

Orolia2s / oscillatord

Daemon used to discipline the Atomic Reference Time Card from Orolia
GNU General Public License v2.0
21 stars 14 forks source link

Survey in error is not updated if GNSS signal is lost #122

Open leoleovich opened 9 months ago

leoleovich commented 9 months ago

Hi guys, We noticed an issue when a GNSS antenna signal is instantly lost (disconnected or died) we update the number of satellites, but not the survey in error which stays on the previous level forever, giving an impression that everything is fine:

06:22:11 DEBUG GNSS data: Fix no fix (1), Fix ok: False, satellites num 0, survey in error: 1.07, antenna status: 4......

Ideally, the survey in should be updated to reflect the issue. Thank you!

leoleovich commented 8 months ago

@agagniere

agagniere commented 8 months ago

Including an excerpt from Ublox documentation for reference (emphasis mine) :

3.1.5.4.1 Survey-in Survey-in is a procedure that is carried out prior to entering Time mode. It estimates the receiver position by building a weighted mean of all valid 3D position solutions. Two major parameters are required when configuring:

  • A minimum observation time defines the minimum observation time independent of the actual number of fixes used for the position estimate. Values can range from one day for high accuracy requirements to a few minutes for coarse position determination.
  • A 3D position standard deviation defines a limit on the spread of positions that contribute to the calculated mean.

Survey-in ends when both requirements are successfully met. The Survey-in status can be queried using the UBX-TIM-SVIN message. The timing receiver should not be fed RTCM corrections while it is in survey-in mode.

So it is normal for the suvery-in error to not change once the survey-in is done, even if a fix is lost afterwards.

Now, if you need an estimate of the accuracy of the position, it is completely possible, we could expose :

Let me know what is your need

leoleovich commented 8 months ago

I guess my question is more on the lines of - can we restart the survey if we lost all satellites? Seems like survey is not very accurate in this event anyway.

agagniere commented 8 months ago

You want to restart it because you want to rely on the survey-in error as being representative of a current state ? Why not use the position/time accuracy indicators for that instead, they are there for that. The survey-in is not meant for this purpose.

leoleovich commented 8 months ago

Need to double-check if it’s expired via the monitoring interface…