How to prevent magnetomer interference from motor wiring
I have something that I wish to share with you guys, I will later incorporate it in the first post.
I was looking for a way to minimize the interference to the magnetometer from the motor wiring when motors are armed and running.
I was pleasantly surprised when I managed to minimize the deflection of the compass to 2 degrees from motors disconnected to all 6 motors running at full power
, and that was using 6 1100KV motors with 8x4.5 props on a 4s Lipo.
Before I made these modification, if I was watching the compass on APM Planner while I Armed the motors and applied a bit of throttle the compass would swing as much as up to 10 degrees and it varied depending on how much throttle was applied or how motors were being controlled.
I am In the process of modifying my frame to make it easier to test various flight controllers, I wanted to mount a plate with shock isolating mountings so that I could then mount a flight controller to that shock isolating plate.
My plate is a single sided unetched fiberglass PCB about 1.5 mm thick with the copper side facing down. It happened to be handy.
The shock isolating grommets I used are Farnell part number 142-2144 GEL MEC - GVD-1010-40 - DAMPER, 10MM
Originally I also used the 142-2160 bolts that are specially designed for these however I quickly discovered they are magnetic and they do affect the magnetometer so I replaced them with a nylon screw and 2 nylon washer carefully tightened so as not to compress the grommet.
First thing I did was to pack some foam rubber on top of my motor wiring [my soldered wires power distribution] so that the wires are pushed down toward the bottom plate and as far away from the magnetometer above as possible
Bottom plate in place - my earlier, smaller food container depicted
Bottom plate removed, foam packed above motor wiring
Lid of my food container attached to top plate with double sided tape, 4 nylon stand offs go through the food contained and attach to the top plate. Hole cut in lid of food container to allow wiring and also air atmospheric pressure to enter the food container, a small hole drilled on the other half of the food container so that it does not become pressurized and give false barometer readings. This food container has a soft blue rubber seal abound the edges. Note I only connect a single Signal wire to each ESC, GND connection is via the power connection from Lipo to the dedicated FC UBEC. See Notes at bottom of this post.
My shock isolated plate [unetched PCB copper on other side], this photo depicts the original bolts which did affect the magnetometer. 4 nylon stand offs to hold the flight controller.
This is NOT the supplied GPS, I was testing a CRIUS CN-06 V2 U-blox GPS module+ Ceramic antenna with my copper PCB GPS antenna ground plane. The magnetic bolts have been replaced with nylon bolts and nylon washers.
This gives you an idea of the distance between top plate and FC board with magnetometer.
The food container I am using has handy locking tabs, it is a bit larger than required but there is plenty of room to test Flight Controllers.
All that work paid off, it took off nice, smooth and level on default PIDs no drama at all, only required a bit of I tweaking, sweet.
You can see I do not have much level grass areas here on the side of a hill.
I am not sure if the fact that I am using an unetched PCB makes any difference, the copper layer would reflect electrical interference [it is NOT grounded].
I suspect that what makes this work so well is the extra distance between the motor wiring and the magnetometer, this works really well.
1.- I am using a dedicated UBEC to power the FC
The UBEC output is set to 6V [jumper selectable between 5V and 6V] and is connected to (-) and (+) on the D46 header of the HK Mega board. On The HK Mega this arrangement supplies 4.8V to all the 5V headers which is well within spec, 3.3V headers are at 3.3V
2. - I am ONLY connecting the signal wire to each ESC
. GND and UBEC power in ESC are NOT connected to FC. this is to minimize ground loops, using a dedicated UBEC only costs less than $5 but it may save you a lot more if the UBEC in the one ESC you decided to use fails and you crash/destroy your toy. This also means there are fewer stiff wires to transfer vibration to the FC.