MKIV Rudder Pedals losing calibration

[InfiniteChimp]

Been playing a bit of Star Wars Squadrons lately and after about 10-20 minutes, I notice my rolls are much easier in one direction than the other. Examining the VKB axes through VKBDEVConfig shows the center-point shifts 30% in one or another direction. Recalibrating fixes the issue temporarily, once again after 10-20 minutes, the problem recurrs.
What is causing this?

Thanks!

[Ron61]

Show the Profile-Axes tab (Log & Phyz axes) two times - after the first calibration and 30 minutes after the second, when the axis floats.

[InfiniteChimp]

Don't mind the X/Y Axis on this photo, I have the stick springless for use in helicopter simulation, it is not presenting any problems. Only the Z axis is having issue (pedals).

[Ron61]

In the first screenshot I see only one calibrated axis - these are the pedals. How so?
The second is better. Try setting MPL=9 for the 7th axis, Set and calibrate. Show the screenshot.
You can put DzLo=3, and the presence of 0 on the first 3 axes in this parameter is surprising.

[InfiniteChimp]

Not sure what the indicator is for a calibrated stick. My GF stick (the one the pedals are attached to) is hanging loose in the forward position, which is why the value is not zeroed out. Also, I set the DzLo to 0 myself, I was assuming this is the deadzone setting and eliminated it for more consistent performance crossing the 0,0 position (as I use the stick mainly as a helicopter cyclic, and it rarely sits at 0,0).

[InfiniteChimp]

*Edit*

I just realized what you asked. At your request I posted a screenshot containing my calibration settings before and after float. Instead of calibrating again and then waiting for 30 minutes to demonstrate the float a 3rd time, I simply took a screenshot of the post-float conditions at the moment I posted here in the forum (as I had just recently calibrated about an hour beforehand, and then subsequently experienced the float) and then calibrated only the pedals to show their relative offset. I did not calibrate the stick (x/y axis) for the screenshot.

[InfiniteChimp]

MP has been set to 9, these are the updated values.

[Ron61]

Now the values of KdHi and KdLo are closer to the recommended values for the Mars sensor, which is in the pedals. If the axis floats away, you will calibrate and show a new screenshot for comparison.

[InfiniteChimp]

Thank you. I will update you if/when this happens again.