Here’s a first attempt to explore the I2’s instability whilst hovering.  Any help, or (polite) suggestions, would be very welcome!!

  

My plan is to try to tie the problem down by exploring the parameters available to view in the DAT files which are recorded during each  flight.   There is a huge amount of information in the DAT files  ranging from motor speed to ESC temperatures, RC movements to GPS data, and lots more which I don’t pretend to understand.

  

I've recorded video of the I2 whilst hovering (with no control inputs and in light winds), and provided sync points so that the data from the DAT file can  be set against it.

When viewing the video the I2s behaviour will be correlated with the events shown by the timeline marker on the DAT data.  The sync points between the video and the DAT data are the two yaw inputs (rotating the I2 through 180 degrees), which can be viewed on the video and seen on the DAT file timeline, along with other parameters of interest.  

  

To get the ball rolling here is a first video -

  

The craft was set up about 4m agl, beyond ground effect, with a camera trained on it.  The I2 was set to record also.  The I2 was rotated 180 by yaw input – this is represented by the blue trace on the screen, the falling edge of which is used as the start sync point. The rising edge of the second yaw input is the second sync. The video images are aligned with this data, so that the movement of the craft should be fairly closely associated with the other events chosen for illustration from the DAT file.  I have picked  Number of GPS sats (red) and Number of Glonass Sats (green).  The steps show the loss or addition of a satellite - Glonass 10 -12, and GPS 11-12.

  

The timeline indicator runs between the sync points.  So the events on the graph should be occurring  in sync with the video and the timeline indicator for the DAT view.

I don't see any definite correllation between I2 movements and GPS signal changes, but I do see the I2 video moving around as the craft sways about!  Not what you want to see really!

  

need to try without Glonass maybe ?
but we cant turn if off...

Is there really a problem here? The movement of the I2 shown in the left video seems OK to me. The FC is making the corrections necessary to keep the I2 in place.  The right video is fixed on a point at infinity just as it should be.

Rather than look at the numSats data you could also look at the navHealth data (I assume youre using DatCon to get the data). You could provide the .DAT via Dropbox, or similar, so that others can take a look.

This Inspire's specific issue was discussed to death on every related forum with no solution. Not every Inspire exhibits such behaviour, some are wobbling just slightly, some wildly, most probably aren't that bad at all (mine including). It's a result of specific motor alignment design engineers decided to employ here, unfortunately ignoring the requirements of film makers in particular applications. In addition, the efficiency and quality variables of GPS/Compass/IMU sensors are playing a deciding role maintaining Inspire's hover stability. No FW version is able to eradicate this dreaded Inspire's wobble completely. In contrast, Mavic or M600 series doesn't suffer such, as employing different motors alignment concept. Needless to say, it's a Achilles heel of Inspire 2 ...

They are better off making a massive Mavic 2 that holds an X7.

But it must be the firmware. Because before the firmware .100 we didn't have the problem with two Inspire 2. With the firmware .200 it got even worse and from the firmware .300 it got better again, but never as good as before the version .100. We also have an Inspire 1 that stands without problems in the air.

DJI can solve the problem if you want. At the moment it looks as if DJI has no time or interest for it. It's funny that no DJI staff member says anything about this problem. It's just dead silent until maybe an Inspire 3 comes out.