No-moving-parts field mill?

This experiment attempts to use ionizing radiation from two americium pellets from smoke detectors to continuously discharge a sense antenna for electric field measurements. As long as the ionization can conduct significantly more current than the bias and leakage current of the buffer amplifier the wire should stay "neutral" and be able to sense a static DC field, or so goes the theory. The circuit is trivial, simply a voltage-follower op-amp (TLC27L7) running on 9 volts connected to the sense wire through a 1 megohm just for safety (let's call it an "idiot resistor"). A 4.5 volt "artificial ground" made with two 1 megohm resistors biases the can halfway up. The high-value resistors and micro-power op-amp will run on a single battery for years.

I'm using the micro-power mosfet op-amp, the TLC27L7, simply because I have them and many CMOS types will work fine. The two americium elements support about 20 pA so that bias current of CMOS op-amps should be plenty low.


I mounted two smoke detector americium elements in an almond can with a pickup wire above them near the opening. This time there is no bias voltage on the can except for the artificial ground; the ions simply "connect" the op-amp to the sky's field by preventing charge accumulation. It's mounted under an awning pointed somewhat towards the ground but it still seems to work. The original idea was this to be the DC portion of a field mill with the faster stuff being sensed by an ac-coupled circuit and another sense wire. But it's pretty quick, as-is, and a frequency response network of some sort might be all that's needed. The real challenge would be to make one that weather and insects don't ravage.

The data is being recorded by my live SID recorder; ignore the red trace.  I'm using my Meter DAQ Amplifier to boost and offset the voltages for recording.

The first plot shows a flat spot near the end is when I covered the open end of the can with a metal lid. A modest weather system has produced more noise and an overall higher reading. A cold front can be seen arriving at 11:00 just before sunrise on the second plot:

I positioned a metal plate hooked to a variable power supply in front of the sensor and was able to plot "DC" lines for hours. For a more serious attempt to make this work well, I'd use a DDS to measure the frequency response and then tailor the gain of the amplifier to compensate. I say a DDS because the frequency response is likely to be changing at very low frequency with the DC response being less than the AC response even below one hertz.

A rainstorm came through with no lightning and made the plot below (left). I copied the next day's data and added it to the previous plot, hence the overlapping top black trace. Yesterday starts at just before 5:00 and the clouds begin arriving in the early morning, around 14:00. Rain starts to fall at 19:30 (sudden rise). There may have been a little electrical activity around 0:00. I needed a good old lightning storm!  I got the storm right before 9:00 UTC (plot to right) the next day. It clearly responds to the change in potential due to a thunderstorm and I'm sure a faster plot would be interesting.

Ideas From This Project

The technique shows promise for other "non-contact" potential measurements but it might be a bit fragile to leave outdoors. I'm thinking of other more robust sources of ionization to try. One idea is the exhaust for our gas water heater. Even the pilot light produces fantastic quantities of ions and perhaps enough of them make the trip up the flue to the roof. Just mount the sense wire in weatherproof metal cans above the flue cover pointed off at an angle towards the ground, one can for the sense wire and a smaller can on the back for the op-amp. (Field mills are usually pointed at the ground to sense the induced charge rather than up, simply to keep the rain out.)

A small dedicated flame, either propane or natural gas, shouldn't cost very much and experiments could be done with an ordinary candle. Just mount the whole thing under an elevated metal trash can pointed at the ground. The sensor would be above the flame near the bottom of the can but I'd orient everything so that the fumes from the flame don't go directly into the chamber. The smallest flame will dwarf the ion production of the americium. With a little thought, the whole thing could probably fit under a bucket.