Electronic Voice Phenomena is one method currently employed in paranormal research. However there is much debate over the best methods of obtaining results. Some say that this phenomena is the result of an EM Field being detected, and not sound at all. There are studies which seem to support this. So how can we validate this theory?
The Type of Inductive Sensor
Many supporting this theory claim that only inductive microphones should be used since they contain a coil instead of capacitive plates. The belief is that an EM Field can interact with the coil to create the EVP signal. But I find two problems with this concept. First, any microphone by nature is designed to pick up sound. Even an inductive microphone is built this way. It contains a diaphragm which moves the coil in a magnetic field when sound is detected thus creating a voltage representing the sound. Consequently, if a signal is detected using an inductive microphone one cannot be certain if it originated as an EM Field or was simply an audible signal picked up by the microphone. Secondly, since the intent of the microphone is to detect sound the coil is shielded against stray fields which otherwise might cause interference. While this shielding may not be perfect it does somewhat impede the effects any EM Fields might have on the microphone. The sensitivity to such fields is reduced by design.
But there is a better method. That is the use of a sensor designed specifically to capture EM Fields in the voice frequencies present in the audible spectrum. A sensor such as this is open to such fields since no shielding against them is provided. Plus, since no diaphragm is used the sensor is immune to any sounds such as talking or someone's gurgling stomach. This article will look at the basics of designing such a sensor.
At first glance it seems all that one needs is a coil of wire which connects to an amplifier. Indeed that is essentially what an inductive sensor is. But how many turns of wire, how large should it be, what sort of core should it have, how can the gain be increased, and what effect will all this have on the ultimate outcome? For that we need to go into the design of the coil. This paper will in no means cover all the details. For that one would need to study AC electrical theory. Entire books have been written on that subject! But hopefully this will give enough of the basics to point you in the right direction if you want to actually design and build a sensor array.
To begin with, I am going to dispense with the math wherever possible. But it can't be entirely omitted. We will cover inductive reactance, capacitive reactance, impedance, resonance, and resistance. There is no way to design a sensor without considering the effect each of these has on the other in a coil.
The value Inductive coils is measured in henrys. Sensors we use for EVP will be smaller, millihenrys or even microhenrys in some cases. Generally the greater the value of the coil the more sensitive it will be to EM Fields when used as a sensor. The value of a coil can be increased by either adding more turns or using an iron or ferrite core. Larger coils or cores will increase the value in henrys. But there are some downsides to simply adding turns, more about that shortly.
Unlike a simple resistor, when an AC signal is applied to an inductor, it effective resistance goes up as frequency is increased. The formula for that is:
This becomes important since we are concerned with a wide range of frequencies over the audible spectrum. Clearly if one plugs in the value of 100 (the low end of the voice band) to F and does the calculation, the results will be considerably different than if one uses 3,000 (The high end of speech). The inductive reactance will be much higher at the higher frequency. The greater reactance should produce a stronger signal as frequency is increased. And this is so, but only to a point. Enter capacitive reactance.
Generally capacitive reactance applies to capacitors, and one wonders why it also is a factor with inductors. But there is capacitance in an inductor; it is based on the surface area of each turn as it lays beside or over an adjacent turn. And the effect is cumulative; every turn adds to the total value. It too can be calculated using the following formula:
You may note the similarity in these two equations. But the important factor here is that capacitive reactance is the reciprocal of inductive reactance. In other words, unlike the inductive reactance as the frequency increases the capacitive reactance will decrease. This tends to limit the high frequency response of our sensor as frequency increases the loading effect on the coil. Thus we can't continue to add turns forever to increase sensitivity.
Increasing the Gain
There is a way we can boost the sensitivity while minimally affecting the capacitive reactance. Up till now we have assumed an air core, that is a coil wound on a form suspended in air. But if we add a ferrite or powdered iron core we can increase the inductance without greatly affecting capacitance since the number of turns remains unchanged. We do however have to choose carefully when selecting cores.
Iron cores are good at low frequencies. But as frequency increases eddy currents develop in the core of any iron material. These adversely affect operation by lowering the gain. However the increased gain of iron at low audio, up to about 1-2 kHz allows its use without serious effects. They would be ideal, especially if one is dealing with low frequencies. This can be a positive too as any stray RF that finds its way in is likely lost due to the eddy currents it produces.
For the full audio spectrum consider ferrite. Ferrite consists several variations of iron compounds There are different grades based on their composition. The most common soft ferrites are Manganese-Zinc (MnZn) and Nickle-Zinc (NiZn). Manganeze-Zinc ferrite will cover anything up to about 1 mHz. which includes the audio range and quite a bit above. Nickle Zinc, while it could be used is not quite as desirable because of its lower density but could extend the response well into RF spectrum. Not really needed for EVP work, and besides, it would easily pass any stray RF signals!
There is an optimum point when the value of capacitive and inductive reactance equal each other. That is the point the circuit becomes resonant. Gain at this point would be extremely high, and if we were building a radio receiver we would aim for this point. Our sensor would become a tuned circuit, passing that one frequency extremely well. But that's not the goal of the inductive sensor, in order to cover a broad range of frequencies we must cover the entire voice band of frequencies, not just one. The frequency of resonance can be determined using the formula:
Some may have noted in doing the calculations that the ratio of inductance to capacitance between turns is very high. In other words millihenrys of inductance to picofarads of capacitance. This is true in our case since we have made no effort to add capacitance as we would have if we were intentionally building a resonant circuit. The inductive sensor is by nature highly weighted toward inductance. But still we cannot ignore the effects as they do become important especially as more turns are added to the coil.
The "Q" of the Coil
Any resonant circuit has a value referred to as "Q" or Quality. This term applies more to radio use since a high Q is a sharper, better defined frequency bandwidth. If we had only inductance and capacitance to deal with the "Q" would be perfect. Only a single frequency would pass through. But that is not ever the case, there are losses and DC resistance of the wire in the coil to consider. The natural point of resonance will cover a narrow band of frequencies. Exactly how narrow is a matter of the coil itself, how tightly or loosely wound it is, and even the core used.
So What Does All This Mean?
To begin with as stated earlier we want wide bandwidth first, then the most gain we can obtain while maintaining that bandwidth. If one looks at a resonance curve one can see a typical bell curve with the peak at resonance. There are two things to consider here. First we must attempt to avoid the peak since the bandwidth is very low. Thus we should design our sensor so its natural resonant frequency is far outside the desired operating region. For most audio sensors I would recommend keeping resonance at least 50 - 75 kHz or higher. That ensures we operate in a more linear region of the response curve.
Next we can decrease the "Q" of the coil by placing a resistor across its output. The lower the value of resistor the wider the bandwidth. But the downside here is that as we load the coil we also reduce its sensitivity. So it kind of becomes a tradeoff, how much gain are we willing to give up to increase the bandwidth?
We can use a ferrite core which will maximize the sensitivity while keeping the turn count low. This has the effect of minimizing capacitive effects between turns and keeps the resonant point higher and away from the range of frequencies we are interested in.
Finally we have the last issue to contend with, that being frequency response. Even across the audio spectrum we will find as frequency increases the coil becomes more sensitive. This is partially offset by the nature of speech itself, the lower frequencies are generally stronger than the higher. But still it can be an issue. The solution here is not so much in the coil but what the coil is connected to. The amplifier should be high impedance to minimize loading effects. In addition it should be frequency compensated; that is built to give a higher gain as frequency decreases. Ideally you can use an audio frequency generator and oscilloscope on your sensor to determine its response curve then pad the first preamp stage to compensate. With a little work this should allow you to get a relatively linear response curve from your sensor.
Tri-Axial Sensor Array
Up to now we have discussed a single coil. In operation as an EVP sensor three coils are recommended, one on each axis, X, Y, and Z. It is not recommended to simply connect three sensors to one input across each other as the loading effect of one sensor on another could be a problem. Nor should they be connected in series as the inductance of the other two will decrease the signal from the first. Instead the solution is a single, frequency compensated preamplifier on each coil, tuned to that particular coil. The output of these preamps should be a low impedance. This, fed to a high impedance mixer will minimize any loading effects. They are then summed together and isolated using the mixing stage and sent on to your recorder or main amplifier. This method will prevent the loading effect and keep the array more stable. Of course all these amplifiers should be supplied from a well regulated, filtered supply and also be totally shielded against any stray RF or EM Fields. A double wall Faraday shield with single point grounding is recommended as well as RF suppression filters on all input, output, and power connections.
Hopefully this article will get you going on the next step in EVP research. Now you can put down the cheap digital voice recorder and join the cutting edge of EVP work.
Dale has raised an issue regarding resonance. He says the by allowing resonance to occur the overall sensitivity can be increased greatly. Furthermore he recommended loading the coil to lower the "Q" of the coil thereby increasing its frequency range.
I addressed this briefly above. However since this issue has been raised I have added a more in depth explanation why I have avoided the use of resonance as a means of increasing the sensitivity. Since of course resonance is a property that cannot be eliminated I have recommended designs which place this property well outside the frequency range being observed.
Figure 1 illustrates the response as I have recommended it. The resonant point here is about 75 kHz which places it well above the range of voice frequencies which comprise EVP. The area shown between the red lines on the graphs represent approximately 20 Hz - 20 kHz. The curve represents the relative sensitivity of the coil at all frequencies between 10 Hz to 100 kHz and above. The peak at about 75 kHz represents resonance, the point where maximum sensitivity is attained.
But more important than over all gain is the frequency response in the region where voice frequencies occur. In figure 1 this point occurs along a relatively linear portion of the curve. The entire dynamic range is shown by the green bar to the right of the graph. This is the signal strength at each of those frequencies within the voice band. Under ideal conditions this green block should be small, representing a constant output over a wide frequency range. It should also be located as close the top of the graph as possible since this represents improved sensitivity. External amplification can do the same thing as raising the overall level of the block. Frequency. compensation can also be done to the signal to give the same effect as making the block smaller. Both of these operations are usually done externally in the preamp which boosts the signal from the sensor.
Figure 2 shows the effects of lowering the resonant point to about 8 kHz, which is within the range of frequencies we are detecting. While certainly the output level is much better it must be noted this only occurs at a narrow range of frequencies near 8 kHz. Frequencies both above and below that are much lower in amplitude as represented by the large green block. This indicates an extreme dynamic range for which we must compensate in the preamp. Considerable filtering and limiting might accomplish this but at a cost. Such a preamp would involve multiple pole filtering and would likely introduce unnecessary noise into the circuit. Thus, unless a particular project requires only a narrow frequency range be monitored, a resonant peak becomes an issue to be avoided.
Regarding the use of a load resistor Dale is correct, the peak amplitude at the resonant point can be lowered by loading the coil. Figure 3 represents the effects of a load resistor across the coil and a resonant point of about 8 kHz. Clearly the dynamic range over the voice band of frequencies has been reduced as is evident by the smaller green block than that seen in Figure 2. Lowering the value of the resistor will provide more load and increase the bandwidth of the coil. Thus the dynamic range between frequencies will be decreased requiring less external compensation.
But notice too that the overall position of the block has been lowered. This is because the load does not just effect the resonant point, it lowers the overall sensitivity of the coil at all frequencies. And the more loading applied the lower the sensitivity since the load diverts signal from the amplifier where we need it. This means that to attain the original gain additional amplification will be needed. More amplification generally increases the background noise from the amplifier stages. Signal to noise ratio will suffer, additional background hiss likely will increase.
For the reasons stated here I have recommended compromising the overall sensitivity slightly and keeping the resonant point outside the range of frequencies being studied. This is the most effective way to attain a wide band coil and still have adequate signal levels to drive a minimally compensated preamp.