Infrasound Project
The contents of this page are copyrighted by the Curious Cryptid, all rights reserved.
The information is made freely available for research purposes in the practice of open science.
That makes it available to you, it does not make it yours; link and/or cite if you want to use the information.
This is an ongoing project started 2020-05-14, last updated then to create the page.
Purpose and Justification
Breakdown of the Project
Infrasound Background Research
Audio Equipment Research
High Level System Design
High Level System Verification Design
Purpose and Justification
It has been conjectured Sasquatch can use infrasound at levels capable of incpapcitating humans for defense. Detecting this would be beneficial to demonstrating if this is so.
An example of this experience is R1H, the lead investigator for our investigative region 1 in the foothils around Mt Ranier. They were camping in a motor-home RV when his wife said something big was moving in the woods near the camper. RlH rushed out the door with a camera in hand and just a few steps from the vehicle suddenly began vomit uncontrollably. The effect lasted for several (as in 12) hours until the nausea passed completely.
From this description alone I would consider infrasound to be a candidate explanation.
From this experience alone I consider the project justified to make a practical infrasound detector for field use. If I am to be potentially exposed to this sort of phenomenon then I am absolutely interested in understanding it better.
Project Breakdown
Infrasound is generated in low levels from many sources in the environment. I do not want the wind blowing through the trees to set it off.
I only want the sensor to provide an alert if the energy of the signal is in the range that is likely to begin affecting humans.
Presenting the data is problematic. I suppose a visual indicator is one possibility. The idea came to me that frequency-shifting the sound may be a better idea. For example to re-present the detected infrasound from 1Hz to 20Hz in an audible range from 100Hz to 2000Hz. This output could easily be piped into an audio recorder for later analysis.
The frequency range affecting humans is of greatest interest. Extremely low frequencies not known to affect us are beyond the required scope for this project. Also clearly audible frequencies are beyond the required scope for this project.
In order to verify the sensor I will need to have a source of infrasound.
Methods to sense the signal also require evaluation.
If the project is successful, exploring a product is in order.
Infrasound Background Research
I do not plan to make a deeply detailed study or report on the effects of infrasound on human subjects. My intent for this project is merely to collect enough data to make a reasonable design of the project to sense infrasound capable of having an effect on humans, and a method to test that sensor.
I am mainly using an NIH study1 on the toxicology of infrasound as my principle reference, unless more is needed.
Infrasound affects humans by physical jostling of organs. The study notes a variety of symptoms for levels of infrasound above aroubd 110db. I was surprised at the level of infrasound from wind and from moving objects. This may require me to rethink how to deal with a sound level at which to trigger an alert.
The paper also observes that due to the long wavelength of infrasound waves, a point source is relatively ineffective for transmission by air because the energy is dispersed fairly quickly. This would suggest it is mainly effective at relatively short distances.
Stiumulation of the vagus nerve leads to a variety of conditions including rapid changes in blood pressure and symptoms related to that.
The majority of human physiological and psychological effects are observed in the range of 6Hz to 20Hz. This would be the minimal required range I want to be able to measure. As an arbitrary target I would aim for 1Hz to 60Hz.
Due to the nature of the required data processing this means sampling at least 120Hz for at least 2 seconds for each data point; that requires a minimum of 240 data points, not a problem. To have multiple updates per second (no more than 10) I would process overlapping data windows. For better resolution I would want to collect more samples, so both a higher frequency and a longer period are options for data collection. Nither memory nor processing speed will be a significant problem here.
Audio Equipment Research
Well, it doesn't take much research to find that both mocrophones and speakers tend to drop off below around 30Hz. There isn't much to be done with microphones at all. Only the largest speakers around 21 inches diameter reach down near 20Hz, but they can be modified.
I know there are infrasound sensor apps for your phones. Technically microphones that cutoff around 20Hz do respond to lower frequencies to some extent. However the loss of energy becomes exponentially greater as the designed frequency is exceeded. This makes them fundamentally unreliable because manufacturing differences, even per unit in the same lot, will likely lead to different responsiveness per unit. It is far better to find something that is simply responsive at lower freqnencies.
Pressure transducers seem to be the best bet for a sensor. I found a study2 citing the use of such a sensor with some degree of success. A low-pressure high-speed transducer, probably like that used in a CPAP machine will probably do nicely with appropriate mechanical considerations.
Speakers can be modified by adding mass and/or by operating them against a sealed box. Normally speakers operate in a vented box so that when the cone moves the air is displaced more or less as if the speaker is in open air. A closed box would reduce the frequency response by forcing the speaker to compress/decompress the air in the enclosure. Similarly a mass added where the coil connects with the cone will reduce the frequency response of the speaker.
High Level System Design
For a proof of concept project there are a broad range of embedded systems available around which the appropriate sensors and software can be written. Given the need for FFT processing I will probably make use of an ESP32, Raspberry Pi, or Orange Pi platform.
Basic principles of diginal signal processing require sampling at least twice the highest frequency I want to resolve, practical experience suggests 4 or more times. If my upper cutoff is 20Hz then I want to sample at least 80Hz; I will start in theory with a sample rate of 128Hz.
Using an FFT to identify energy by frequency instead of time I will need to sample an even power of 2 number of samples, say 256 samples. This should give me resolution of 1/2Hz in the output.
That would mean it takes 2 seconds to collect 1 frame of data. I would like this to update roughly like a digital multimeter (~4Hz) so I will need to process 4 frames per second, these frames will overlap, but that is OK. Once I collect the first 256 samples, I will process a frame every 32 samples.
I will discard results clearly in the audible range, say 40Hz and above. I should be able to present a spectrum from 0.5Hz to 40Hz with resolution of 0.5Hz provided my sensor is responsive enough.
With each update I would simply re-process the truncated FFT data into a (100x) higher frequency range to produce an audible output from 50Hz to 4kHz with tones updating 4 times per second.
High Level System Verification Design
A separate embedded system will be needed to produce infrasound signals to test the sensor effectively. For proof of concept the accuracy of energy transfer is not terribly important. The generation of an accurate frequency is more important.
