Help needed: Reducing EMI with a metal detector coil mounted under a rover

[monkey32]

Hello everyone,
We are Matchbox Tech, a small team developing SandWorm, an experimental autonomous metal-detecting rover.
The project is designed to let the rover handle repetitive walking and systematic scanning while the detectorist remains in control of the search and digging. More information about the project is available here:

https://matchbox-tech.com/

OUR CURRENT PROBLEM
The search coil is mounted underneath the rover. Because of the limited ground clearance, the coil is relatively close to the chassis, motors, motor drivers and high-current wiring.
We are still seeing unwanted electromagnetic interference and unstable detector readings when the rover is operating. We suspect the problem may involve both electrical interference from the motors and controllers, as well as eddy-current effects from metal parts near the coil.
WHAT WE HAVE TRIED
- Repositioning the coil away from the motors and high-current wiring
- Reducing the amount of metal near the sensing area
- Modifying the chassis structure
- Adding shielding to several electronic components
- Improving the separation between the detector and the vehicle electronics
These changes reduced the interference, but they did not eliminate it completely.
We would appreciate advice on the following:
1. What is the best way to shield the coil and its cable without reducing detection sensitivity or detuning the coil?
2. Would ferrite cores, common-mode chokes, motor suppression capacitors or additional filtering on the motor wiring help?
3. What grounding arrangement is recommended for the detector, motor drivers and battery system?
4. How can we determine whether the remaining noise is radiated EMI, conducted noise or interference caused by the metal chassis?
5. Has anyone here successfully mounted a VLF or PI detector on a moving vehicle or rover?
SandWorm is still a Gen 2 experimental prototype. We are currently continuing mobility and mixed-terrain testing, and we are sharing both our progress and the problems we encounter.
We can provide more details about the detector type, operating frequency, motor voltage, wiring layout and before/after test results if needed.
Any practical advice or examples from your own projects would be greatly appreciated.
Thank you!

[boilcoil]

Hello monkey32,
All your suspicions regarding electromagnetic interference and eddy currents are correct.
Here is what I would suggest:
1. The issue of electromagnetic interference should be addressed not by adding extra shielding to the coil itself, but by reducing the level of emissions from the motors, controllers, cables, etc. It is advisable to use shielded power cables for the motors and controllers and to ground them at a single point on the rover's chassis. Motors are generally shielded because they have a metal housing connected to the chassis, but it is good practice to ensure a solid connection to the chassis.
From the photo, I see that you have a wireless connection and GPS for coordinate logging; these could also affect the performance of a highly sensitive detector. You could mitigate their impact on the detector by placing shielding between them and the detector unit.
2. To achieve optimal detection results, you need to remove—or at least minimize—any metal objects near the search coil. Closed conductive loops within the range of the coil's search field are particularly problematic; they generate strong eddy currents and, consequently, a strong magnetic field of their own. This invariably degrades the detector's performance.
However, in my opinion, your main problem is interference from the motor power supply, and your primary efforts should be directed towards reducing radiated electromagnetic noise.

[Byko]

- H-field from a small loop falls off as 1/r³: doubling the coil-to-source distance cuts the noise by about 18 dB (8× in amplitude).
- Twist all motor and supply wires; use a twisted triple for 3-phase motors.
- Mount the drivers at the motors to keep the switched wires as short as possible.
- Keep the motor PWM well above the detector band (e.g. 40 kHz PWM with a 6–8 kHz detector). Also keep it off odd harmonics of the TX frequency (e.g. 5 × 8 kHz = 40 kHz), since square-wave synchronous demodulators respond at 3f, 5f and 7f.
- Keep all ferrous material, including shields and the motors themselves, away from the coil.
- Enclose the drivers in a magnetic or conductive shield. Steel gives the best H-field suppression, 2 mm aluminum (skin depth about 0.58 mm at 20 kHz) is a good non-ferrous alternative. Filter all cables at the enclosure exit with common-mode chokes (2-3 turns).

[Carl-NC]

First, what is the metal detector design? VLF, MF, or PI? Frequency?
Second, have you disconnected the coil to see if the noise is from H-field coupling into the coil, or E-field coupling into the circuitry?
Third, if you run the rover around on pavement is it quiet or still noisy?

I've never built a rover but I have built a walk-through, and then parked an X-ray machine right up against the coil panel. Overwhelmingly, the biggest noise source is from the motor that drives the conveyor belt, and not from the X-ray imager itself. With walk-throughs, we use interference-canceling coils, almost always butterfly coils. They are good at canceling single point sources of magnetic EMI 90° to the panel. In your case, I assume you have multiple stepper motors as magnetic EMI sources so a coaxial or top-hat coil might do a better job of suppression. A coaxial/top-hat coil will also help suppress on-edge detection of metal, i.e., from the rover itself.

Assuming the drive motors are the primary culprit, another thing to try is to synchronize the TX frequency to the motor frequency, and also make sure all the motor frequencies are all synchronized. That's also something we do with walk-throughs: synchronize the TX frequency to the AC mains.

[pito]

show the picture of the electronics where they are located, the black frame = it is a metal ?

[Carl-NC]

Frame is carbon fiber.