Could someone test compass calibration of the external compass with last copter 3.5.4 and last mission planner? I’ve just try with 3 different pixracer they all have really absurd results. Using MP wizard with the “old globe” visualization 3 different Pixracer show a calibration done in a “2d line”, many times happening (this 2d path tracing) in a strange out of center position…but really I have so weird results.
Interesting the first time I’ve tryed with the third pixracer, the first compass calibration seem to works ok, drawing a correct 3d path around the globe, making a second one brings as the other 2 different pixracer build acts so wrong 
…after some more testing I’ve found something interesting!
I’ve restored to default all the parameters and the compass calibration on the wizard works again, made the calibration and MP says: “setting new offsets for compass #1 failed” as soon as this happen MP sets compass_DIAX around 442 or more (default 1). And this compass_DIAX its the cause of the weird compass calibration, the issue remain because I cannot made a succefull calibration, because also after resetting campass_DIAX to 1, after a good calibration MP says “setting new offsets for compass #1 failed” and so the weird tunnel happen again.
Is this happening to you? please check
P.S.: with an old version 1.3.43 this problem is not present !
Use onboard calibration.