June 2006, Issue 191

Earth Field Magnetometer
Cypress PSoC High Integration Challenge 2004 Winner

CONSTRUCTION

I initially cannibalized a car compass so I could use the sensor along with its drive circuitry. The original sense amplifier was fine for a slow-moving motorized compass indicator, but it was too noisy for this application. So, I ended up building a new amplifier by following the original design with a few minor changes. Placing it farther away from the noisy drive circuitry and powering it from a separate regulated supply made it much quieter. As you can see on the right side of the main board in Photo 1, I used the original drive PCB (without the sense amplifier components).

When I went to buy another compass to use as a spare, I was dismayed to find out that it was no longer available. All of the low-cost electronic compasses available today use a newer type of variable inductance sensor made by Precision Navigation. It’s much simpler and smaller than a fluxgate for compass applications, but it isn’t as straightforward for accurate magnitude measurement. Fluxgate sensors are still used in marine and aviation systems, but they’re well out of my price range. I realized it was time to learn how to build my own sensor.

My first task was to find an appropriate toroid core. The ideal core material should have a sharp saturation characteristic. This is known as a square loop, which refers to the shape of its B-H curve. It should also be a tape-wound core made up of numerous thin layers of a material (e.g., 4-79 Moly-Permalloy). I settled for a readily available ferrite core, although I didn’t expect it to be quite as sensitive. The core is a Magnetics ZH42206TC toroid core, which is a highly permeable material (µ_R = 15,000) that makes it easy to saturate.

Photo 3 shows my homegrown sensor. I wound 42 turns of 20-AWG magnet wire around the core. The exact number of turns wasn’t critical, but I knew the winding should be symmetrical. I used just enough wire to completely cover the inside of the core with a single layer. I ended up with an inductance of 16 mH, although this value wasn’t critical either. The core was then sandwiched between a pair of 1² square covers from miniature plastic boxes. I used foam tape to hold the assembly together. I then wound 500 turns of 36-AWG wire around the outside for the sense coil.

I built the Helmholtz coil out of a short piece of 3² ID plastic drainpipe and some sliced sections of pipe coupling that I slid over it. The coupling pieces were spaced apart to leave room for the windings, which are 217 turns each. I used 28-AWG wire, which just filled up the available space. The fluxgate sensor was mounted with nonferrous hardware in the center of the pipe.

I wire-wrapped the rest of the system. I was careful to keep the noisy circuitry away from the sensor input. The band-pass filter has a rather high voltage gain of 208, so I was careful to keep its input and output separated as much as possible.

The rest was simple. I mounted everything on an acrylic sheet. To minimize the amount of noise, I placed the sensor assembly a few inches away from the electronics. I put the whole system inside a large Styrofoam container because the sensor is sensitive to temperature variations.