Infrasound Monitoring with a Microbarograph


How to detect low frequency acoustic waves in the atmosphere.

This article appeared in Volume 5, Number 4 of the Bell Jar.


A host of pressure disturbances are created in and are propagated through the environment of the earth. Seismic activity is one of the better known and a significant number of amateurs are engaged in monitoring the waves produced by earthquakes and volcanic eruptions.

At sites that are good for seismic observing even relatively small earthquakes can be registered throughout the world by fairly simple apparatus. However, it is found that it is only the lower frequency components of the earthquake wave that propagate to the greater distances. This is because absorption decreases with decreasing frequency and the dominant losses at low frequencies are geometric. Most wave phenomena share this characteristic [1].

Just as the low frequency components of seismic waves spread through the earth from their points of origin, so can acoustic pressure waves travel long distances through the atmosphere. These far-travelling disturbances occupy the frequency range below that of human hearing and are therefore termed infrasound. Horizontal propagation of these waves occurs in sound ducts created by temperature gradients and is modified by the winds of the upper atmosphere with winds at altitudes of 40 to 60 km having the greatest influence [2].

Infrasound can be generated by natural causes such as severe weather, frontal passages, meteoric fireballs, the aurora, mountain associated waves and avalanches. Human activity also can generate infrasound. Examples include aircraft, large machinery and explosions. Taken together, these have been termed the “atmospheric wave zoo.”

While somewhat differing definitions exist, we will consider the infrasound spectrum to extend from about 0.001 Hz to a few Hz. At higher frequencies, propagation distance is limited and there is considerable noise from the din of the local environment. Below this frequency range we have the long timescale, large amplitude atmospheric pressure variations that are due to normal meteorological processes. This is the effect that one observes on a standard barometer.

So, where do these infrasonic signals fit into scientific research? Much of the activity on infrasound began during the Cold War. Nuclear detonations generate infrasonic signatures and arrays of detectors can assess the yield and location of an explosion. Today there are plans to implement a worldwide network of detectors that will be used in conjunction with seismic sensors to test compliance to the Comprehensive Nuclear Test Ban Treaty. Activities of a less political nature include the study of the origins and propagation of acoustic gravity waves (longitudinal pressure waves in the atmosphere, not to be confused with the elusive gravity waves predicted by general relativity), the detection of large meteor-fireballs (bolides) that intersect with the earth, and research into the coupling between the upper and lower parts of our atmosphere.

For the amateur, there are several areas that may be worth pursuing. For example, it might be interesting to include an infrasound observation capability with a seismology program. A challenge would be to try to correlate the passage of acoustic gravity waves with reported meteorological conditions. An observing program might also involve studying possible relationships between acoustic waves in the lower atmosphere and internal gravity waves in the ionosphere. The latter are transverse waves that may be detected by observing the propagation of low to medium frequency radio waves [3]. Both can be produced by disturbances in the atmosphere. Infrasound from the explosive disintegration of bolides could be verified by checking a database of visually observed fireballs such as the one maintained by the International Meteor Organization’s Fireball Data Center. Several interesting papers on meteor infrasound have been published [4, 5, 6]. A number of topics are discussed in a recent broad review of infrasound [7].


The acoustic pressure sensors that are used to detect these low level, low frequency waves are called microbarographs. Basically, a microbarograph is nothing more than a microphone that has been optimized for high sensitivity at very low frequencies. These sensors also have to be designed to minimize pressure effects that are not of interest.

Seismic sensors have an advantage in that they get bolted to solid earth and they only move when a disturbance hits them. The microbarograph has to detect very low pressure fluctuations of periods ranging from a fraction of a second to minutes that ride on top of the constantly changing absolute pressure (the barometric pressure) of the atmosphere. Barometric pressure variations tend to have much longer periods than those of the infrasound signals so this effect may be removed by incorporating a high pass filter element into the microbarograph.

This is done by using a sensitive differential pressure sensor with one side of the sensor pretty much open to the atmosphere and the other side connected to a finite and stable volume which is connected to the atmosphere by means of a low conductance element: a leak. Thus, while one side of the sensor can respond quickly, the other has a “leaky volume” that gives the device a high pass characteristic. The sizing of the leak and the reference volume will determine the time constant of the device, hence the lower limit of the frequency response. Figure 1 shows the general configuration.

With atmospheric pressure variations out of the way, the next step is to tune the instrument to the area of interest. Infrasound occupies a broad range of frequencies and pressure levels and, as luck would have it, no one instrument will adequately cover the entire spectrum. Dividing the spectrum into two bands is typical. A higher passband, say from about 0.05 to 10 Hz is useful for detecting close and relatively small events. This is the range that is used by the defense establishment to listen for low yield atmospheric nuclear weapons testing. A lower passband, from about 0.003 to 0.03 Hz is generally used for monitoring natural infrasound emissions and acoustic gravity waves. In either case, sensitivities are typically on the order of 0.1 to 1.0 microbar and events may have amplitudes to 10s of microbars. (A microbar is defined as a pressure of 1 dyne per cm2 and is equivalent to 0.75 milliTorr or 0.75 micron Hg.)

Verification and localization of events is essential in serious work and this requires a network of microbarographs. The network might consist of several microbarographs located on a grid covering a few square kilometers or it might be a complex global network. For any amateur who is familiar with seismology, this will all sound familiar. For the detection of acoustic gravity waves resulting from distant disruptive events, the specifications of the sensors of a French microbarograph network [8] are instructive. The passband is 0.003 to 0.04 Hz (periods of 333 sec to 26 sec) with a sensitivity of 1 microbar. With instruments deployed in such locations as the Ivory Coast, French Polynesia and, of course, France, this network recorded and localized the explosion of Mount St. Helens from distances as far as 11,500 km. Peak amplitudes were on the order of +/- 50 microbar.


A number of sensor technologies have been used for microbarographs. Professional instruments typically use modified high quality capacitance microphones. These have sensors that determine pressure changes by measuring the displacement of a diaphragm by electrical means. However, with price tags in the multikilobuck range, this was not an option for me.

When undertaking this project I looked at two specific types of differential transducer: solid state pressure sensors as exemplified by Honeywell Micro Switch's 163PC01D36 (±3.5 in H2O [about 6.5 Torr] range), and capacitance diaphragm gauges (CDGs) as exemplified by MKS Instruments’ Model 223 Baratron® (±200 milliTorr range). The former costs about $100, the latter about $465.

I found the Honeywell device to have an electronic noise level roughly equivalent to 10 milliTorr peak to peak whereas the CDG's noise was not detectable. Also, the solid state device tended to have a rather high zero drift with nominal ambient temperature changes, nearly 100x the CDG's drift. Finally, the CDG has an accuracy of 0.5% of full scale and a resolution of 0.02 milliTorr, parameters that are not specified for the solid state sensor.

Given the sensitivity and stability differences, plus being a long time vacuum enthusiast, the CDG seemed to be a natural route. Normally used on process equipment, these work something like capacitance microphones but would be lousy for normal audio applications because their response drops off above about 10 Hz. No problem for this application.

This is not to say that the Honeywell device wouldn’t produce some interesting results. A few weeks ago I got a call from a Chicagoan who detected the sonic boom of the shuttle entering the atmosphere using such a sensor.

The diagram of Figure 1 shows how the CDG is integrated into the microbarograph. The sensor consists of a welded capsule which is divided by a tensioned Inconel diaphragm. Behind the diaphragm is a ceramic substrate upon which there is a metal pattern in a form similar to that of a circular bull’s eye surrounded by a ring. As a result of this geometry, between each electrode and the diaphragm there is a capacitance that is defined by the areas of each electrode, the dielectric constant of the intervening gas (1 for air or vacuum), and the separation of each electrode from the diaphragm. Each of these variable capacitors is a leg in a bridge circuit. When the diaphragm is flat, the bridge is balanced and there is an output of 0 volts. However, as the diaphragm becomes curved, the bridge is unbalanced and the degree of imbalance is reflected in the output voltage. The conversion from voltage to pressure is set by the manufacturer when the device is calibrated against a pressure standard.

Each CDG is specified according to its full scale range and it will be able to measure pressure for some number of decades below that full scale. In the case of my 200 milliTorr device, the output will be precisely 1 volt at 200 milliTorr and the voltage will vary linearly with pressure. Thus 20 milliTorr will result in a 0.1 volt output, and so forth. Needless to say, as the pressure goes to, say, a hundredth of a percent of the full scale, the output voltage will be pretty small. This is about the bottom end of where a useful reading can be obtained and will correspond to the resolution of the device.

Every CDG requires that the electronics be zeroed to balance the bridge. For a differential gauge all you have to do is connect the ports (to assure the same pressure) and, with a voltmeter connected to the terminals, turn the zero pot until the output is zero volts.

As noted before, in the microbarograph configuration the reference port of the CDG is connected to a reference volume. That volume, in turn, is connected to the atmosphere. With the leak adjusted properly, long, slow variations in atmospheric pressure will affect both sides of the CDG’s diaphragm equally. Shorter period variations will upset the equilibrium since the reference volume will not respond quickly enough to prevent a differential pressure from developing across the diaphragm. Thus, the time constant that is represented by the interplay between the size of the reference volume and the conductance of the leak is what sets the low frequency response of the microbarograph.

Since we are dealing with very low differential pressures, temperature effects will have a significant effect on the instrument. A very slight warming of the reference volume will cause the air to expand, increasing the pressure and leading to a false pressure indication. In a practical installation, the entire assembly must be housed in an isothermal container where short term temperature variations are minimal.

My implementation is shown in Figure 2.

The CDG is connected to this manifold by short lengths of plastic tubing. The manifold is constructed from 1/8-inch brass plumbing fittings. The reference volume consists of this manifold, the reference side of the CDG and the connecting tubing. To permit the volume to be varied, I added a plastic syringe. Once I get the device tuned properly, this will be replaced with a fixed volume made from plastic pipe and fittings. Two cheap brass needle valves (also from the hardware store) are incorporated to allow the CDG to be zeroed without having to disconnect it from the manifold. Whenever adjustments are to be made in the plumbing, the valve between the CDG’s ports should be open to prevent overpressure conditions from developing across the diaphragm. (More on this below.) A sintered metal diffuser of the type used on air tools is used to keep dirt and dust out of the manifold and CDG.

One option for the leak would be another needle valve. However, I found it more convenient to use hypodermic tubing (available in 6-inch lengths from Small Parts Inc., 13980 N.W. 58th Court, P.O. Box 4650, Miami Lakes, FL 33014-0650). Several leaks were made by gluing various diameters and lengths of hypodermic tubing (0.005 to 0.008 inch inside diameter) to threaded brass fittings using epoxy cement. The response of the microbarograph may then be adjusted in a repeatable fashion by varying the volume of the syringe and the size of the hypodermic tubing.

The time constant can be determined with coarse accuracy by connecting a scope to the CDG and then creating a slightly negative pressure on the reference side by slightly withdrawing the plunger of the syringe. The scope will show the characteristic exponential curve as the pressure comes back to equilibrium. The time constant and, therefore, the frequency turnover can then be determined.

Note that I specified a negative pressure on the reference side. It is best to avoid positive overpressures on the reference side of some differential CDGs as the diaphragm can become distorted when it stretches toward the baffle. Usually the measurement side can take positive pressures to 20 psi or more because the diaphragm is not harmed by flattening against the ceramic/metal electrode structure. One configuration of the MKS Type 223 is available with 20 psi overpressure on either side and that is the configuration that should be specified if going this route.

Figure 3 shows the completed microbarograph.

I mounted all of the components on a piece of plywood that was sized to fit snugly into a “Playmate” cooler. The cooler helps to minimize fast temperature variations. The lid fits securely enough to keep rain out but, since it is not air tight, the instrument works properly. The electronics package is a simple fixed gain amplifier made from two op-amps. This is used to boost the output of the CDG to a level that is adequate to drive the recording system.

Proper siting of the instrument is important. Keep it outdoors, away from heavy traffic, out of the sun and shielded from the wind. Wooded areas are good locations. The instrument and its insulated box may also be located indoors with a length of moderate diameter tube (e.g. a piece of garden hose) leading out of doors to an appropriate area. The French microbarographs of Reference 8 used a 3 meter length of 1.5 cm tubing for this purpose.

At this point, I have completed the basic instrument and have run some tests with the instrument indoors, near an open door. The next steps will be to site the instrument in the woods behind the house and optimize the frequency response.

While I haven’t recorded anything of great significance, the device is very sensitive. Moving a door anywhere in the house will show a strong response. Figure 4 shows the instrument with a “slow” response. The left hand side of the trace shows the exponential response of the reference volume as the pressure equalizes. The three oscillations at the center were produced when I slowly swung a closet door at the other end of the house.

Figure 5 shows a recording made at higher sensitivity and with a faster response (time constant about 1 sec). The signal toward the right hand side of the trace corresponds to the passage of a jet airliner as it made its approach to the Manchester, NH airport, about 15 miles distant. I can't say for sure if the trace is due to the jet. However, nothing else seemed to be going on at the moment.

Next Steps

Two items are on the plate as next steps. The first is to set up and configure the computer based data recording and analysis system. The second is to add a noise reduction system to the device. For data recording I am using two programs that were developed for amateur seismology by Larry Cochrane, the driving force behind the Redwood City Public Seismic Network. The first program, Seismic Data Recorder (SDR), is a DOS program that is used to collect the data. It is used in conjunction with a PC Labs 711s AtoD card (available from Jameco, (800) 831-4242) or a similar (and less expensive) card that Larry produces. The second program, WinQuake, is used to view and analyze the recorded events. Both of these programs are free and are available via the PSN's Web site

With regard to the second item, I will be looking at ways to decrease the effects of local noise. A variety of noise reduction techniques have been developed to minimize the effects of wind noise. Simply baffling the inlet with a piece of foam (like the wind shields seen on microphones) will have some positive effect but the best performance is obtained with arrays of long lengths of perforated or porous hoses that are connected to the inlet of the microbarograph. References 7, 9 and 10 may consulted for further information.

Finally, it must be noted that I only looked at two commercially available pressure transducers. This should not prevent the amateur from evaluating other units or even attempting to build the sensing element.


I would like to thank Bruce Kendall of Penn State University who first brought this topic to my attention, Rodney Whitaker of Los Alamos National Laboratory who was kind enough to provide useful comments as this article was being developed, and Peter D. Hingley, Librarian of the Royal Astronomical Society, who provided reprints of several sections from Reference 7. I would also like to extend appreciation to MKS Instruments, Inc. who provided the capacitance manometer that was used in this project.


1. Rodney W. Whitaker, "Infrasonic Monitoring," Phillips Lab Monitoring Symposium, 1995.

2. Earle E. Gossard and William H. Hooke, "Waves in the Atmosphere," Elsevier Science Publishers, 1975, Chapter 9.

3. Jearl Walker, "Billows in the Ionosphere are tracked with transistor radios" in The Amateur Scientist, Scientific American, September, 1980.

4. Douglas O. ReVelle, "Historical Detection of Atmospheric Impacts by Large Bolides Using Acoustic-Gravity Waves," International Conference on Near-Earth Objects, Sandia National Laboratory, April 24-26, 1995.

5. Douglas O. ReVelle, "Infrasonic Observations of Meteors," 1995 Monitoring Technologies Conference, May 15-18, 1995.

6. Douglas O. ReVelle, "On Meteor Generated Infrasound," Journal of Geophysical Research, Vol. 81, 1217-1237, 1976.

7. The Geophysical Journal of the Royal Astronomical Society, Vol. 26. (This is an issue devoted to infrasound propagation theory, instrumentation and observation.)

8. C. Delclos, E. Blanc, P. Broche, F. Glangeaud and J.L. Lacoume, "Processing and Interpretations of Microbarograph Signals Generated by the Explosion of Mount St. Helens," Journal of Geophysical Research, Vol. 95, D5, 5485 (April 20, 1990).

9. F.B. Daniels, "Noise Reducing Line Microphone for Frequencies Below 1 cps," Journal of the Acoustical Society of America, Vol. 31, 529-531, 1959.

10. S.D. Noel and W.W. Whitaker, "Comparisons of Noise Reduction Systems," Los Alamos report LA-12008-MS, 1991.

Home     tBJ Articles & Publications    Chambers & Supplies    Contact tBJ
©2008 Stephen P Hansen, the Bell Jar