PPM - New Design, Tuning-Free

The old concept, that is used in almost all proton magnetometers, consists at least out of two analogue filter stages: The input circuit, made out of the polarization coils, is made resonant by a bank of switched capacitors. Due to the higher Q-factor around the resonance frequency, a first signal amplification is achieved, comparable to what happens in a simple radio receiver. After the first stage a tuneable bandpass filter (e.g. in biquad technology) follows, the center frequency and the Q-factor can be adjusted by the use of digital potentiometers, controlled by the microcontroller. Finally, after amplification, the zero crossings are detected with a comparator. The edges are used to sample the value of a counter in the microcontroller into a register for further processing.

In the ages of digital signal processing analogue filters look pretty ugly and the frequency determination by just using a comparator is not optimal too. Actual 32-bit microcontrollers offer fast floating point arithmetic, large RAM sizes and integrated 12-bit ADCs. Therefore a controller of the STM32-family was used in the new magnetometer, because the old ATmega with its 8-bit architecture was way to slow for complex signal processing as it is needed here. The STM32 is also a good device for hobbyists, many information sources are available on the Internet too.

Many experts think that the resonant coils circuit is especially, precious and hard to replace by digital processing because it seems to deliver an easy and noise-free amplification. However, the price is a higher source-impedance, that results in larger noise due to the current-noise of the input of the first amplifier. Therefore the removal of the resonant coils turned out to be not as dramatic a expected. But nevertheless a further reduction of the noise in the first stage (now around 1nV/rt(Hz)) was required. The simplest way for improvement was the simplification to a single-ended input. This now does not allow cable lengths of 10 meters but that is not a real disadvantage in the field far from urban buildings and electrical lines. The following figure shows a simplified block diagram of the new design:

The pair of coils forming the sensor (1) is either connected to the battery voltage by a mosfet switch (3) or to the amplifier (6). The current limiter (4) ensures that the electronics is not damaged if the cable is broken and shorted. A reverse-voltage protection is not needed anymore thanks to the use of protected voltage regulators. The simplified design of the signal conditioning (no filters) now allows a simpler single supply (5). The noise if the system results in only a fluctuation of +-1nT in magnetically clean environments, with a measuring period of 1s.

The mechanical construction was done with the case from Fischer, that was used in very early designs too. Thanks to the significantly higher processor power a graphics display could be used, controlled via SPI After the power-on the measurements immediately start, no tuning or so is needed. The possible settings were reduced by intention and include just the selection of the measuring period (1-3s) and the definition of an alarm threshold, when crossed a high-efficient LED is flashed on the upper right corner. The black block below is a lead-acid accumulator.

The device is not using any storage of data or settings which makes the usage as simple as a voltage meter, just power on and see the first reading after seconds. For further processing of the measurement values a bluetooth-module is built-in which allows the storage of date on a smartphone. For this purpose an app was written, that allows to display every measured magnetic field value on a map and in a trace window in realtime. The scaling of the trace can be changed with finger-zoom, it also defines the value range of the color scale:

A smartphone can generated disturbances and the usual built-in GPS is not really optimal, also the readability of the display in direct sunlight is often a problem. However, the usage of a smartphone is so attractive that there is no real alternative to it. Almost unlimited memory, high-res display, map rendering possible and much more. To achieve the same with other means (Raspberry, HDMI-display...) would be a much higher effort, difficult procurement of needed parts, software development and finally higher costs too.

A further step is possible when the display is moved to a smartphone too: The "packaging" of the sensor together with the electronics and the battery in ab backpack. Also this results in further disturbances, but the are limited to the one-digit nT range and there is hope that they can be further reduced by other optimisations (complete removal of the display, less current consumption, less current through the battery and therefore less fields generated). With a distance piece the following assembly can be used. Outside the backpack only the smartphone is required:

Here comes a short look into the circuit cellar where the coils were winded with a self-made machine that was driven by a DC-motor out of an old IT tap drive:

I think this will bring my work on the proton magnetometer to an end at the moment. The proof of the concept without the tuning of the input circuit is made. Would be interesting to get the digit right from the comma stable, with a measuring time of 1s. This is clearly possible with Overhauser magnetometers. Unfortunately I could never test a classical proton magnetometer with a period time of 1s. If somebody has access to a proton magnetometer from one of the professional manufacturers, I'd be happy to get in touch.

At the end I'd like to mention a very cool page from Graeme Keon. He built a circuit for children to allow listening the protons dancing! I had the idea to present something similar but Graeme was faster, and it is really nice:
http://motivationtolearn.org/wordpress/?p=1765 (seems to be offline at the moment)

Now I almost forgot to mention that some things have been found with my proton magnetometer. Just one example below. I am not an expert for such stuff but it is pretty impressive: