Magfield problems

Hello, good afternoon, how are you?
I have a fleet of drones that use GPS:
DP0503 RTK GNSS (XL M8P)
I usually use them flying in groups, and I always had problems with magfield, for example, in fly with 30 drones, about 20 accuse magfield when I turn on.
Recently analyzing magfield and magfield2 data, I noticed that I have a huge variation in the data, for example magfield1 which is the GPS, starts at 190~200, but when I move the drone to any direction, it goes up to 800, but then that I do compass it sets in the range of 200 which is a good number according to ardupilot wiki.
However, if I turn off the same drone that was calibrated and had a good value, sometimes just by turning it on, it comes back with instability and accuses magfield.
I “cleaned” as much of the cables and components as possible that could be interfering with the FC and gps, and even then, it’s still bad.
I needed to improve this because in a show of 60, 70 drones, if we have to do 30 bars, we lose almost all the drums and there’s nothing left to do the show.
Could someone help me with this problem, please? I will be forever grateful for the answers!
I don’t know if it’s the quality of the GPS, or some parameter that doesn’t store its offsets and generates this magfield, I even raised the hypothesis that it’s the territory I’m in (Brazil), because I have contact with friends from other places in the world and they don’t have this problem, using the same materials/drones.

I think it’s important to notify is that when I rotate the drone, the GPS magfield variation is very large, for example, at one point it reaches 170 and at another, 500, 550, just by rotating the drone.

Regarding the “fitness” that appears on the MissionPlanner compass page, does it influence anything on the GPS accuracy?