Testing heading from dual gps


I am trying to test dual gps for yaw, however I am uncertain for what to look for in order to confirm that it is actually working.

I am following this setup:

When I have been testing the second gps achieves rtk-lock and the first achieves 3d-lock.

GLOBAL_POSITION_INT/hdg seems to drift somewhat, and the behavior of this field seems to be unchanged when I disconnect the second gps. This leads me to believe that this field in on itself is not a good place to look in order to determine whether the gps-heading works. (or that it is an indication that it doesn’t work).

As I understand it this field is the result of a fusion between the IMU data and gps-data using ekf3.
Is there any way I can see what the pixhawk calculates as heading from the gpses, without it being fused with other sensor-data? What I wish to achieve with this is testing the whether (and how well) the gps-heading works independently, or whether ekf3 is just using other sensors to estimate the heading.

I am using two zed-f9 gpses, and pixhawk 4. One gps is connected to the gps-port and the second is connected to telem2. The two gpses are placed 60 cm apart on a rod which i use for testing.
I am running copter version 4.0.5

When I rotate the pixhawk the heading seems to be updated correspondingly, but I believe this might be due to the IMU.
When I only rotate the rod with the gspes, the hdg seems to be unaffected. Does this imply that the setup necessarily doesn’t work, or might it be due to the fact that it is fused with other data?

My question is what I should look for to confirm definitely that the dual-gps heading works well.

I am new to ardupilot and any help would be greatly appreciated! Thank you

It is nice that we have something in common. I did not dig much in debugging the source code for “printing out” those values but having some tricks to check the situation.

You may want to check with the magnet for disturbing the compass while connecting 2 gpses. If the dual-gps works, heading (yaw value in Mission Planner) should remain unchange.

If there is no compass in your system, trying to shake the autopilot will also generate disturbance because of gyro drifting. Therefore yaw value will increase or decrease.

Thank you.

I had a successful test today. It appears the reason I had trouble using hdg to test whether the pixhawk used dual gps for yaw was that the gps fell out of rtk-fix all the time.
When I tested now with stable rtk-fix, and turned off the magnetometer, it was clear to see that the pixhawk had a stable heading when the gps was connected, and started to drift when I disconneted it.
Also when I changed the direction the pixhawk was pointing without changing the position of the gps the heading changed slightly, but then returned to the direction the gpses were pointing. This was sufficient to conclude that the dual heading indeed works.