EFNMR stands for Earth Field Nuclear Magnetic Resonance and is a fancy way of saying proton magnetometer. I undertook this project because I was fascinated by the prospect of measuring the movement of an atom.

1. Idea
2. Schematic
   1. Analog Front End
3. Current Situation

---

Idea

Magnetic nuclear resonance is a fascinating topic and relies on the principle that particles with non-zero nuclear spin align themselves within a magnetic field. This physical phenomena is utilized for a large variety of machines and sensing technology. In many cases it requires a very strong and uniform magnetic field in order to make the different particle perturbation frequencies sufficiently far apart for identification.

For this project I utilize the same principle; however, instead of relying on a strong magnetic field I rely on the earth’s magnetic field. This field in very small areas is incredible uniform and we can use this to measure a proton’s precession. Overall I designed this machine so that you could create a crude 2D cross-section image of an object that has been immersed in water or oil. The proton precession amplitude is what will be measured.

---

Schematic

Analog Front End

Because the signal that we’re measuring is incredible weak there are a few things I’m doing to make sure we can detect the proton precession. In order of importance:

1. Tuned LC tank. In order to selectively hear the 1.5-2kHz signal we tune our coil using an LC tank circuit. This automatically helps create a high Q narrow band pass filter of the region of interest.
2. High Gain amplification. For this I’m using an AD8428 fixed 66dB gain low noise instrumentation amplifier. In addition to its low noise characteristics its not exorbitantly expensive like jfets. You can dead-bug these amplifiers to improve the noise characteristics as outlined in this application note. Awesome paper in general is this one.
3. 4th order Chebyshev bandpass filter. After the first amplification stage we go through another stage configured as a fourth order filter. This helps again narrow down the frequency region of interest.
4. Oversample digitization. Because our signal range is 1.5-2kHz we can easily oversample to reduce the noise a step further. Digitization in this case will utilize an audio I2S 32bit 192kHz digitizer.

---

Current Situation

This project has been ongoing for a little bit now. My intial PCB bringup showed that the split rail power supply had issues regarding stability. I’m currently redesigning it to take into account a different strategy.

As for the analog chain I’ve breadboarded the circuit and showed that I can detect very small field fluctuations when the sense coil is coupled with a weak electromagnet.