Effects of altitude on Earth's electric field and the reception of natural very low frequency (VLF) radio events

From Whitworth Near Space
Jump to: navigation, search

Research Group: Nathan Fickert, Chris Grochowski, Jon Hoff, and Ryan Richardson

Launch: Whitworth Fall 2012

A radio receiver in an insulated pod was sent up in a weather balloon to record atmospheric noise, or sferics, in attempt to detect very low frequency (VLF) events. In this experiment, the noise was recorded using a radio receiver connected to an audio recorder. The receiver was designed to detect variations in electric field in the audio range between 150 Hz and 10 kHz. A vertical one meter wire was connected to the input of the receiver and protruded externally from the pod to have maximum exposure to possible VLF events. The receiver was shielded by a metal casing to prevent any interference with the circuit. The data recorded consisted of sferics, short pulses of electromagnetic radiation from lightning. These pulses were integrated over time, giving us measurements of sferics per minute. There was a significant increase in sferic activity between the balloon's liftoff and 1,230 feet, the average activity increasing to roughly 110 sferics per minute. This suggests that a slight increase in altitude from the ground significantly enhances the ability to measure atmospheric noise. The activity remained relatively constant for the remainder of the ascent. Beginning at the descent, the recorded data was diluted with sounds from the pod's thrashing behavior. The sferics per minute settled in the range of 0-8 upon landing.

Background

The earth’s atmosphere is filled with electronic noise that can be detected at higher altitudes. Using an appropriate receiver, these signals can be converted to an audio recording distinguishable to the human ear. A major source of this noise is lightning because the current that flows through the lighting channel acts like a radio antenna. This radiates signals at a broad range of frequencies which propagate through the atmosphere. The most common frequencies found in the atmosphere are ELF (extra low frequency), VLF (very low frequency), and atmospheric noise. VLF and ELF waves are detected as whistlers or cracks because of their impulse-like shape in time.[1] They get their unique whistling quality when the energy from the signal travels through the ionosphere and is scattered by plasma in the magnetosphere.[2] They propagate along field lines across the earth to the opposite hemisphere at frequencies between 3 kHz and 30 kHz.[3] Atmospherics are easily detectable and propagate very well through the atmosphere between the ground and the ionosphere. It has been shown that the best time to detect atmospherics is at night and lowest during midday and during the summer months. This is because the ion density of the atmosphere changes due to the earth’s position with respect to the sun during different times of the day and year.[4]

Mechanical

Fig. 2.1: Interior view of VLF receiver pod showing physical layout.

The circuit was sent to near space in an 18x15x15 cm Styrofoam pod which was covered in metallic foil. On the ascent and descent, the pod was expected to encounter a lot of turbulence from the wind whipping it from side to side or into another pod. To deal with this problem, all of the components were securely fastened to the inside of the pod using wire-ties.

As pictured in Fig. 2.1, the receiver and recorder were secured inside of the pod and saline hand warmers were used to ensure that the inside the pod stayed within the range of reliable operating temperatures for the electrical components.

Electrical

Fig. 3.1: VLF receiver and audio amplifier board in RF-shielded enclosure.
Fig. 3.2: VLF radio: receiver section.[5]
Fig. 3.3: LM386-Based audio amplifier.[6]

The electrical system of our experiment consisted of four main parts:

  • Antenna/field probe
  • VLF receiver
  • Audio amplifier
  • Sound recorder

The antenna consisted of a 1 meter length of very flexible insulated stranded copper wire (the type used for multimeter probes). This wire hung vertically from the bottom of our pod and was soldered to the conductor of a coaxial cable inside the pod. The coaxial cable was approximately half a meter long and was coiled up on the bottom of the pod so that the shielding could act as a sort of counterpoise against which to reference the signal coming in on the antenna. The other end of the coaxial cable was terminated with a phono plug (RCA style) which plugged into the antenna jack on our receiver.

The receiver design comes directly from Stephen P. McGreevy's website on VLF radio reception.[7] The receiver itself consists of three stages. The first is its front end. The antenna is capacitively coupled to Q1 through C1, while C9 shunts excessively high frequency signals directly to ground and also helps prevent unwanted oscillation. Q1 is a junction field-effect transistor (JFET), which has a very high input resistance and uses an electrostatic charge on the "gate" electrode to control current moving from source to drain through the device by varying the amount of available charge carriers. This allows a varying electric field on the antenna to produce a proportionally varying voltage at the output of the device. Although this is a radio receiver, no detection or demodulation is necessary since our desired signal is already in the range of audible sound. All that is necessary is to filter our signal to remove activity outside our frequency range of interest, and this is accomplished by the second stage. The second stage is an inductive-capacitive network consisting of C2, C3, C5, C6, and L1. L1 is one half of the secondary winding of a center-tapped 8 ohm to 1K ohm audio transformer, coupling the output of the JFET to the input of the third stage (which consists of Q2, a 2N3904 bipolar junction transistor biased for Class-A operation by R4, R5, R6, and R7). Our only notable departure from McGreevy's original design is our decision to substitute a MPF-102 JFET[8] in place of the 2N3819 specified in the original schematic. This substitution was made based on availability of parts and should have virtually no bearing on circuit operation or performance.

From the receiver, our filtered and amplified audio signal is capacitively coupled to the potentiometer input of the final amplification stage. The audio output amplifier was based around an LM-386 low-power audio amplifier chip, with external feedback added for a fixed gain of around 11. The overall output level is actually set by the 100k potentiometer. The design was originally intended as a headphone amplifier and is available from the Minidisc.org[9] website. The receiver and the audio amplifier were constructed on a single piece of perforated prototyping board[10] using through-hole soldering techniques. This style of construction gave us a very rugged circuit, which was necessary in order to endure the harsh environment and mechanical vibrations of our balloon flight. Another benefit to this construction was its light weight and small form factor. The circuit layout was designed so as to be small enough to fit inside a mint tin, which served to shield the circuit from any electrical noise caused by the other pods. Both circuits were powered from a single external 9 V lithium battery.

The final step in our signal's path is the digital voice recorder.[11] The output of the receiver was connected to the microphone input of the voice recorder using a cable with 1/8 inch phone plugs at each end.

Software

Our data consist of a single audio file, so all processing and analysis thereof were performed post-flight. In order to process this data, a custom application was written using Native Instruments’ “Reaktor” software, which is a visual programming language (similar to LabView) for designing digital signal processing (DSP) software instruments. The program is intended primarily for use in music and audio production, but its access to low-level functions allows for an the design of virtual software instruments which can perform almost any function or operation on an audio file. We needed to construct a custom software instrument which could perform the following tasks:

  • Play back our captured audio.
  • Use frequency-based filtering and amplitude gating to separate our desired signal (sferics) from background noise.
  • Perform a time-based analysis on the relative levels of sferic and noise activity in a given window of time.
  • Ensure that any undesired noise signals are not counted among the sferic activity.
  • Determine a weighted value for sferic activity given the net amount of time analyzed (and not muted by noise reduction) in a window.


Fig. 4.1: High-level structure of analysis program.

Signal flow:

An audio file is loaded into sample player and audio is then split into two paths: one for data (sferics) and one for noise reduction. Each signal is then run through an equalizer--the low frequency component is removed from the sferics audio and the higher frequencies are removed from the noise signal. The two streams are then sent to two separately triggered audio gates. Upon reaching a gate, the incoming signal is split again, this time into a data channel and a trigger source. The data stream is sent to the audio input and the trigger stream is sent through another equalizer and then to the “sidechain” input, used to control the flow of audio through that gate. This setup allows us to fine tune the trigger for each of the gates. For the sferic gate, the trigger is looking in the region of about 8 kHz, and no sound come through unless the gate detects activity in the 8kHz region. When the amplitude of the trigger signal in this region is above a set threshold, the gate opens and allows the data through. This ensures that our audio output is entirely silent except in the presence of actual sferic pulses. The noise stream is processed similarly, and its trigger frequency is set in the hundreds of Hz.

Front panel of software-based audio analysis instrument.
Fig. 4.2: "Data Analyzer" module.

From the gates, the two signal paths are recombined in the form of audible output that allows the user to listen to the behavior of the application and fine-tune each setting by ear. At the same time, the sferic stream and the noise stream are sent on to another level of the program for numerical analysis. The heart of our data analysis module is the window timer, which consists of a 1 kHz clock and various counters. This module outputs a pulse once every 60 second window, providing timing for the other analysis modules. Additionally, it outputs an integer value corresponding to the current time interval.

Fig. 4.3: "Window Timer" module.

The sferic counter module takes the incoming sferic audio data (which varies between -1 and 1), rectifies it so that all values are positive, and samples the amplitude of the rectified signal 1000 times per second. It keeps adding the value of each sample to the sum of all previous samples. When it receives the appropriate signal from the window timer, it outputs the value of the sum of all sferics from that window and resets itself to zero.

Fig. 4.4: "Sferic Counter" module.

The noise count module monitors the activity of the noise-sensing gate. Anytime there is noise due to low frequency rumbling (caused primarily by mechanical noise such as our pod antenna acting as a microphone when smacking an adjacent pod), the gate opens and sends a signal to the sferic analysis module. This disables the input of that module so that it stops adding to its count when noise is present, helping to ensure that only our desired sferic signals are counted. At the same time, the master clock signal from the window timer is applied to a counter. This keeps a tally of how much time in a given window consists of unusable data.

Fig. 4.5: "Noise Counter" module.

The numbers from the sferic count and the noise count are sent to a module that performs a weighted analysis of sferic activity for that window. It subtracts the amount of time the noise reduction muted the sferic counter from the overall length of the window, divides this value by the window length, and multiplies the inverse of the result by the overall relative sferic count for that window. For example, if the input is muted for 30 out of 60 seconds, the overall sferic count will be multiplied by two in order to give a better idea of the average activity in that window.

Fig. 4.6: "Weighted Sferic Activity" module.

Finally, audio data is sent to a spectrum analyzer to aid in establishing appropriate settings for the gates and equalizers. It is also sent to a recorder that allows for capturing the processed audio, muted in the presence of noise, and with sferics isolated. Running counts of sferic and noise activity for the current window are numerically displayed on the front panel, and the final data output of each 60 second window are captured and displayed in a table.

Testing

The purpose of the testing process was to make sure that the detector and recorder worked as they were supposed to and that they were ready to be sent up to near space. The first testing phase consisted of trying to detect noise generated by scratching a metal wire. The noise that the detector picked up was converted to audio and was able to be heard by plugging earphones (or speakers) into the audio recorder to hear it in real-time. The circuit was also able to detect other electromagnetic noises generated by computers and fluorescent lights. The detector was also tested to see if it was able to pick up any atmospheric noise on ground level. The detector was placed in an open area and left for several hours. The gain of the audio amplifier stage was adjusted so that the maximum output level of the receiver would not clip the audio input of the recorder. Further analysis of the test audio revealed that along with the very loud 60 Hz hum due to nearby power lines, atmospherics were detected and we concluded that the device was ready to be put in the pod and sent to near space

Data and Analysis

The experimental apparatus collected data for the entirety of the flight of the balloon, generating roughly four hours of audio. In order to analyze this data, the audio was processed by the software, which generated a table giving the number of sferics captured each minute for the entirety of the flight. The software also gave the amount of time in each minute that the audio showed signs of physical noise such as bumping, jostling, and whipping of the antennae. By noting the large spike in this type of noise in minute 39 (12:10 PM) of the recording, we could equate this minute with the launch of the pod. With the liftoff time known in our recording, we could compare our data against the flight data generated by both the APRS pod and the command module in order to note patterns in our data corresponding with events in the flight of the pod.

Fig. 6.1: Graph of sferic activity during flight plotted with the amount of physical noise time.

By comparing the audio recording against the altitude readings given by the APRS pod, we noted several key features of sferic measurement. First, as soon as liftoff was achieved, the apparatus immediately began to measure relatively high sferic activity, in the range of 70-150 sferics each minute. This activity lasted from minutes 40-82 (12:11-12:53 PM), generating an average of 110 sferics per minute. During the window of time from minutes 82-87 (12:53-12:58 PM), sferic activity drops into the range of 25-50 events per minute. This marked decline is also accompanied by a spike in physical noise, suggesting that the decline in sferic activity is due to a large amount of physical noise interfering with the ability of the device to successfully record sferics. According to the altitude measurements, the balloon pops at minute 91 (1:02 PM) of the recording. During the descent of the pod to Earth, sferic activity begins to drop off, eventually settling in the range of 0-8 sferics per minute for the time period from minutes 107-135 (1:18-1:46 PM) and until the end of the recording. This data might suggest that sferic activity varies with altitude, or at least that our ability to measure it varies as such. However, at the beginning of the flight, rather than a gradual increase in measured sferics, we see a sudden jump in activity, suggesting instead that there is a relatively low altitude threshold above which sferic activity is much more easily measured. To explain the decrease in sferics with altitude during the descent, then, it should be noted that there is also a large increase in the measured amount of physical noise during this time. During the descent, from minutes 91-122 (1:02-1:33 PM), the average physical noise time is 5905 ms/s, compared with 510 ms/s during the relative calm and high measured sferic activity from minutes 40-82 (12:11-12:53 PM). This high amount of physical noise during the descent interfered with the ability to accurately determine sferic activity during this time period, and so rather than the decline in measured sferics being the result of a decline in altitude, it is a result of an increase in physical noise.

The data shows, then, that the ability to measure sferics is greatly increased by a relatively small gain in altitude. The increased measurement ability does not substantially vary with height in the range from 1,230-20,000 feet, but does rely heavily on the presence of periods of low physical noise. This type of period is best attained on the ascent of balloon, as the descent is a more violent time for the pods, equating to more physical noise. The increase in measurements above a low altitude threshold justifies the search for sferics in near space, as we are able to avoid the apparent interference which close proximity to the Earth provides. Also, we were able to measure average sferic activity for our region and time of 110 sferics per minute, providing a baseline for future research.

While we were able to establish a baseline measurement for average sferic activity, it should be noted that the measurements were taken around noon in the early winter, both of which correspond to the periods of lowest activity in terms of time of day and season. Our measurements should be read a baseline measurement of a period of low-activity, and should not be read as an average in terms of any time of day or year.

References

  1. Holzworth, R. H., M. P. McCarthy, R. F. Pfaff, A. R. Jacobson, W. L. Willcockson, and D. E. Rowland (2011), Lightning‐generated whistler waves observed by probes on the Communication/Navigation Outage Forecast System satellite at low latitudes, J. Geophys. Res., 116, A06306. <http://webflash.ess.washington.edu/publications/Holzworth.cnofsJGR2010JA016198.pdf>
  2. Introduction to Whistler Waves in the Magnetosphere.<http://vlf.stanford.edu/research/introduction-whistler-waves-magnetosphere>
  3. Moore, R C Moore, S. Fujimaru, M. Cohen, M. Gołkowski, and M. J. Mccarrick. "On the Altitude of the ELF/VLF Source Region Generated during "beat-wave" HF Heating Experiments." Geophysical Research Letters (2012). <http://www.researchgate.net/publication/232807184_On_the_altitude_of_the_ELFVLF_source_region_generated_during_beat-wave_HF_heating_experiments>.
  4. Orville, R. E. (1995), "Lightning detection from ground and space", in Volland, H., Handbook of Atmospheric Electrodynamics, I.
  5. Stephen P. McGreevy's BBB-4 (Bare Bones Basic) Natural VLF Radio Receiver Schematic|http://www.auroralchorus.com/bbb4rx3.htm
  6. Design based on Stephen H. Lafferty's "HeadBanger Headphone Amp Construction Kit"|http://www.minidisc.org/headbanger.html
  7. Stephen P. McGreevy's BBB-4 (Bare Bones Basic) Natural VLF Radio Receiver Schematic|http://www.auroralchorus.com/bbb4rx3.htm
  8. MPF102 JFET VHF Amplifier|http://www.onsemi.com/pub_link/Collateral/MPF102-D.PDF
  9. http://www.minidisc.org/headbanger.html
  10. Multipurpose PC Board with 417 Holes|Multipurpose PC Board with 417 Holes
  11. Olympus VN-7100 1GB Digital Voice Recorder|http://www.getolympus.com/us/en/audio/digital-recorders/vn-7100.html