Delay in GPS switchover

Hi,
need some help with some unusual GPS problem.
Recently upgraded to 4.1.0 whilst setting up newly build drone. (Using Navio2 + RPi3B+)
Been using this controller a few years now and lost a drone in June (fatal crash) due to GPS failure followed by EKF issues triggering failsafe mode (default: landing). Which didn’t work well as it happened above a rainforest. :frowning_face:

So, needless to say want to prevent a repeat and decided to install a second GPS on this new build. However, whilst trying to do some tuning I switched over to Loiter mode and at that moment suffered a fly away. Managed to return drone in Stabilize mode.
The .bin file indicates there was a sudden loss of GPS 0 (main GPS). Yet GPS1 was absolutely fine but took around 15 seconds before system switched over to second GPS.
So, I can not answer two questions:

  • Why did GPS0 suddenly fail in the first place?
  • Why did it take so long before system switched to second GPS
  • (Subsequent question would be: Is there a way to make switchover faster?)

Appreciate any help. - thanks

So, if I understand the recent changes / updates to the GPS switchover programming correctly, then the GPS remains set to primary as long as 3D lock is reported ?

Am I correct in assuming if there is a problem in the data stream from the GPS unit, this will not trigger a switchover to second unit as long as there is a 3D lock reported ?

  • This is assuming GPS parameter is set to “Use Best”

…or do I get the whole thing wrong?

I’m trying to work out is the “Blending” option then a better and safer choice rather than expecting a fast switchover in the event the 1st. GPS encounters a problem?

AFAIK these rules are followed:
GPS Auto Switch Options

Thanks Dave,
yep, that pretty much confirms what I’ve read in the Github pages relating to GPS programming in Arducopter.

So it is clear when looking at my recent problem that second GPS appears to be fine at time problem occurred yet system didn’t switch over.

The most likely reasons for not switching over are:

  • A programming issue as mentioned on Github
  • As described in the docs …
    “1: Use best - Automatically select the best GPS, this is done based on GPS fix status (2D / 3D / rtk) if both GPSs have the same fix status the one with the larger number of satellites is used.”

So, in other words we guys who use Navio2 and have the GPS problem we will have limited options as the system won’t switch over whilst a 3D fix is reported. Regardless if an interruption of data flow occurred or data is corrupt. - That is at least the conclusion I draw.

The “blending” function is not a real option either as as the docs state “Blend is best suited for use with two identical GPS units” So in order to have some redundancy one would usually be inclined to use a different GPS unit as the second GPS. :unamused:

So I guess the only option I and others with the same problem have is to change the primary GPS to the one working the best which is most likely the second unit.
(Or start over again by using a different FC altogether)

Once again thanks for your knowledgeable input.

Took a quick look at the log and what i noticed is that GPS0 doesn’t degrade at all, it just stops comunicating with the navio, when it comes back all values are equal or better than when it lost comunication with board. It looks to me like GPS0 was receiving fine all the time it is just the navio that lost it somehow, suspect some serial comunication prob.
5-6 Years ago we were using navio, after 3-4 unesplicable failure events we switched to cube, we have never had a “strange” thing happening since. I would not fly a navio on a drone worth more than 10 Eur :slight_smile:

Thanks for that.
I know all to well about those Navio2 & GPS related glitches. (Unfortunately found out the hard way. Various GPS related weird problems and one fatal crash :unamused: )
There has been suggestions on the Emlid Forum of limiting the amount of satellites the build-in Ublox GPS unit is set to and in that way improving signal performance / reliability. But I still don’t trust that FC anymore.
So in fact I’m currently looking at something like the Pixhawk PX4 or the Durandal option, but haven’t seen many reports about them yet. ( And I don’t like the idea of becoming a “guinea-pig” for yet another FC,…needless to say why)
You don’t happen to have any experience with any of those ?

I guess the main thing right now is to work out the finer details when running two GPS units as I don’t just want to transfer one existing problem to another FC. → I.e. should I even bother with second GPS unit on a new FC?

We use cube orange on a kore board and drotek or ardusimple mini F9P gps with rtk corrections from a cors (either from an on board modem or trough MP telemetry rtk injection). Never had a problem, hundreds hours air time.
I know rtk is overkill but we need it for what we do. F9P it has great performances even if not in RTK.

I moved to CUAV two years ago with CUAV V5+.

I have also a Nora and one X7. Best choice for integration is Nora.

Wow, that’s some serious kit.
Both the cube and the Nora.
One would think it’ll be future proof for the next few years. :+1:

Sorry, somewhat off topic:
Marc, are those the high efficiency foldable props I can see? If so, by how much would you rate it ? 15% or more?

There are KDE foldable props. Efficiency is comparable to RC-Timer but better than DJI (10%).

The great change is vibration level.