LibrePilot Forum

General Category => General Discussion => Topic started by: proctoidpaul on October 13, 2016, 11:11:38 pm

Title: Normal GPS Drift?
Post by: proctoidpaul on October 13, 2016, 11:11:38 pm

OK... admittedly a newbie with Revolution board.

After questionable results with the cheap OP GPS, I ordered a Mini-Ublox-M8N-GPS from Banggood (item # 1035454).   It still shows a huge "drift" while just sitting still even though it has a positive lock on 13-15 satellites.  (See attached photo of GCS screen).

Is this normal?  I am getting very similar actual drift when trying to go into PositionHold mode.

Paul ???

Title: Re: Normal GPS Drift?
Post by: f5soh on October 14, 2016, 12:01:31 am

That not exactly "drift" but most of the time jumps, look at distance between the dots.
After a jump, the position return slowly to the normal position.

Receiving sats below a tree can alter the accuracy.

Take a look here, maybe those changes can reduce the jumps:
https://forum.librepilot.org/index.php?topic=1202.msg15524#msg15524

Title: Re: Normal GPS Drift?
Post by: proctoidpaul on October 14, 2016, 02:19:28 am

Thanks, I'll have a look at the other post.

But from my watching it live, each dot (jump) seems to keep wandering along the way.  EVENTUALLY it seems to hit a point and head back towards the starting point, but certainly not between every jump.  Then it usually never makes it quite back to the right spot again either.

Title: Re: Normal GPS Drift?
Post by: Mateusz on October 14, 2016, 03:39:01 pm

Quote from: proctoidpaul on October 14, 2016, 02:19:28 am

Thanks, I'll have a look at the other post.

But from my watching it live, each dot (jump) seems to keep wandering along the way.  EVENTUALLY it seems to hit a point and head back towards the starting point, but certainly not between every jump.  Then it usually never makes it quite back to the right spot again either.

• Red route is cooked gps value, click on map select "Show dignostics" this will show green route which is your RAW GPS signal. Red route will likely be jumpy because EKF will diverge if it's left without movement for too long.
• GPS signal with jumps 4-6 meters is rather normal, that's why other sensors are calibrated correctly are needed to help GPS. Perhaps better antenna more robust against multi-path reflections would also help.
• Using I2C for magnetometer might be problematic if wires are too long, pull-ups might be too weak. If you have scope around check how magnetometer signal looks like, and if you can improve it by adding resistors. Otherwise, DJI Naze compatible GPS is a better, more robust against noise option.

Title: Re: Normal GPS Drift?
Post by: proctoidpaul on October 14, 2016, 08:14:03 pm

Very helpful!  Thanks a bunch.