One cannot ordinarily record earthquakes via a computer sound card as they don’t respond much to the slow vibrations experienced in earthquakes. (Update: not strictly true. See my latest on this: Simplest Seismometer – Experiments with direct recording through PC sound card) However, with some cheap electronic trickery and clever software, it can be done, and very adequately for the purposes of study. The seismometer above, made by a high school student, in its first night of testing recorded this local 3.8 quake – a magnitude 1 in Taipei:
(click on the pictures for larger versions)
and as a surprise bonus, some filtering of background noise revealed this distant quake as well. It was a 5.9 from 1000 km away in the Philippines.
This blog post is a follow-up to the original post on this subject “Recording Earthquakes for Beginners“. My quest for the simplest and cheapest seismometer design that a student could make, and with a computer, record real earthquakes, has been as successful as I could have hoped. There are software improvements to make one day that will improve the way data is displayed, and experiments are being done on a Windows version, but that’s going to need more study. (Update: The first Windows version has been done. See “Cheap seismometer using sound card and Amaseis – now running under Windows“) The design as it stands has proven quite workable and very educational, plus it can be built by high school students and even younger if they possess some basic handcraft skills. My first student to make one was only in elementary grade four, although he’s very advanced for his age. Following are some photos showing construction details and my students who made this project.
Two channel recording
It turned out to be a very simple matter to also connect the magnetometer described in this post: http://motivationtolearn.org/wordpress/?p=396 It was just a matter of adding two LEDs and a solar cell, then connect it to the second audio channel. This gives us a great way to research “earthquake precursors” – in this case, electromagnetic phenomena which can precede a big quake by days, weeks or even more. You can find more information in this very informative report on current research: http://www.quakefinder.com/research/pdf/Bleier_2010.pdf
The magnetometer traces recorded look much like the following 24 hour recording taken between 00:00 and 24:00 GMT on August 6th, 2011. These were recorded by two different units constructed almost identically and situated about 5 km apart in Taipei. The traces also show a distinct correlation with USGS magnetometers and others.
Construction of the mirror/magnet assembly of the magnetometer is described in the original post at http://motivationtolearn.org/wordpress/?p=396 .
The LEDs and solar cell were attached to a single piece of circuit board as shown here:
And from the other side, you can see that the light is adjusted to fall on the edge of the solar cell – half on. half off. That way we can record changes in either direction. This picture shows the experimental version. It still has the the indicator dial at the far end, no longer used here. A later one was built on a much shorter wooden base.
This drawing shows how it works:
Note: A piece of card was attached beside the LEDs to block light shining sideways onto the solar cell. Also, a box was placed over it to keep out other light sources, and it was lined with black paper to reduce internal reflections. Simon painted all his wooden pieces black as well.
The first thing to do is make the base and arm assembly. The seismometer usually took around 12 hours, but keep in mind that the students who made this were not building a kit set. They did everything, including sawing the wood….
Being a kind of educational experiment, I tried to avoid touching the students’ work myself as I wanted them to learn as much as possible, plus I wanted to find out how much they could do unassisted. I sometimes had to check for accuracy, and there were a few times I had to straighten out some mistakes such as in winding the coil. Safety was also an issue – some close supervision was needed when they used the electric drill and other tools. Other than that, they did practically everything themselves.
The pickup coils were made by hand. This required some concentration since there were a thousand turns of fine copper wire. A pen body was used as a coil former with side pieces from plastic bottle caps glued on. This wasn’t an easy way to make it – later I found the plastic center of a spool of ribbon was about the right size for a coil former. Luckily Shirley didn’t object too much to winding the ribbon around something else.
The picture above shows the piece of lead sheet wrapped around the arm which is simply for weight – about 1 Kg. The magnets, 15mm dia. x 4mm thick neodymium magnets were placed as close as possible to the coil and are adjusted to sit over one side of the windings. The larger damping magnet has a piece of aluminium in front of it. This was glued to a block of wood with brass sides extending down the sides of the frame, which allow it to be slid closer to or further from the magnet, for more or less damping of the arm movement.
A piece of iron wire connects the arm to the turnbuckle, then a piece of guitar string is used to connect it to the top of the frame, where it passes over a small steel plate with a “V” filed in it. The turnbuckle is used for leveling the arm. The screws in the side arms are used for adjusting the angle of the base. These allow us to set the arm so that it hovers about half way between the two wooden stops.
The interface electronics was built inside a small plastic box which now sits on top of the computer: (click to see a larger version)
And here is a circuit diagram: (click to see a larger version)
For those who are already familiar with electronics, here is a brief description of how it works:
The signal from the seismometer coil is amplified by a common LM386 audio amplifier I.C.. It has a gain of 200 with some simple low pass filtering to suppress signals above about 5 Hz. The signal is then chopped at 5 KHz by a simple 555 I.C. so that the sound card will process it as an audio signal. Next, software filters the signal, retrieves the original low frequency signals from the LM386 and stores it in hourly data files. These files can be read by Amaseis, a popular software package used by amateur seismologists.
The magnetometer simply uses the output of the 555 I.C. to drive the LEDs on and off at 5 KHz. The solar cell picks this up as an audio signal and passes it directly to the computer where the software processes and stores it the same way. Two LEDs were used so that the signal would be large and the audio card preamp would not be needed.
It took about 4 hours to make the electronics without a printed circuit board, but by the time they finished, they knew how to solder!
The software currently only runs under Linux. (Update April 20, 2012, a Windows-based, single channel system has been developed here and it displays in real time.) It is open source, written in the C programming language and needs to be properly set up on the computer one is using. It is very small and will run on practically any old computer. I originally set it up on a computer about ten years old, running an old version of Linux that required very little memory. You can download it here: 2chop2-0
When running, it looks like this:
The first column is the sample counter. It will reach 42,000 or so, then save the data (one hour) and start recording the next hour’s data. The next two columns are a drift compensation figure. This is necessary for the seismometer to keep the trace on the zero line, but it has been disabled for the magnetometer so we can see long-term changes. (There is some drift due to temperature and/or voltage changes, but very slight and of no consequence here.) The next two columns are the recorded data. Channel 1 in this case is the magnetometer, channel 2 is the seismometer.
The gain on the audio channel for the magnetometer should be set to give a reading of more than 10,000 when the light is fully on the solar cell. It should then be able to drop back to around -14000 when no light is on it. The unit is then rotated until one gets a reading as close as possible to zero.
To set up the seismometer, set the level control on the interface to maximum, then set the audio gain to just over zero. Use the level control on the interface to back it off to zero, or as close as possible to it. (It’s not easy to get exactly zero, but +/- 100 should be OK. The software drift compensation will do the rest.)
If your sound card or sound driver cannot set the gain of each channel separately, you will probably have to have a fairly high gain setting and use the level control on the interface to reduce the seismometer output. This is OK, it’s just a little harder to adjust it to near zero.
Each hour’s data is stored in folders for year/month/day and they must be transferred to Amaseis’s working directory so they can be viewed. I use a second computer connected via a network, but one can also use Amaseis in Linux if “wine” is installed (software that allows Windows software to run under Linux.)
Contact me via the contact page or leave a comment if you need assistance getting it running.
I will post any improvements to the software here as an update.
UPDATE Sep 19
So far, two of the students’ seismometers have been running for extended periods. Jim reported that his clearly recorded one within the first couple of days, though he hasn’t yet sent me the data files. Timson has recorded some and I have copies of his data files. Here is a nice one he recorded from somewhere off the east coast of Taiwan.
And here is an extract of the report from the local Central Weather Bureau’s website.
Update October 25, 2011
Simon and I continue to monitor and compare magnetometer readings in the hope that we may observe earthquake precursors, reasoning that any unusual signals appearing on both of our detectors and not on Beijing’s, may be such local precursors.
So far we haven’t observed any such signals, however we can see clearly that this system which costs next-to-nothing to make, does record solar phenomena very well. Yesterday’s trace showed an interesting event in the middle of the night, a relatively interference-free period, which allowed us to get some idea of the calibration of our detectors. Beijing shows it as about a 30 nT (nanotesla – a unit of magnetism named after Nicola Tesla) variation. Mine has been scaled down to about half of Simon’s because of the high level of interference at this location, so the highlighted event appears about half the size on mine. It tells us the sensitivity of the two is about the same. It also tells us that the lift in Simon’s building causes a distortion in the magnetic field roughly equal to a 70 nT variation, and the buses passing my house about 300nT. It’s probably a bit ambitious to be looking for earthquake precursors in such circumstances, still it has great potential as an educational project.
Update Nov 9, 2011
Although it was not intended, the magnetometers seem to respond quite well to earthquakes. The mirror and magnet assembly are very sensitive to any physical shaking. Many have been observed to date which were below an intensity of 1 locally, although not distant ones, since the swinging period is a bit short to respond to them. The most distant recorded so far was a 6.9 yesterday which occurred 500 Km away from Taipei. Simon’s magnetometer recorded it as follows:
Zooming in a little and with some appropriate filtering, the moments of arrival of both primary and secondary waves are quite distinct:
My seismometer recorded it similarly, as below:
These were all recorded via the sound card using the software and very simple circuitry described above. If one made the magnetometer by itself, it would be ridiculously cheap to make – the most expensive parts were the solar cell at around US$2.75 and the small bottle of milk at about $1.
Update Dec 19, 2011
The “milk bottle magnetometer” has broken its own record for distant earthquakes with a 7.1 from New Guinea, about 4,000 Kms away from Taipei. The black line is the seismometer trace. The red line is the magnetometer on the other audio channel with some filtering applied (pasted on with graphics software, for comparison.) Only the primary waves are visible on the magnetometer, the secondary waves being too slow to show up on such a short pendulum.
Update Jan 02, 2012
Timson recorded this 6.6 earthquake from Russia on his seismometer:
And this 6.8 from Japan:
He hasn’t been recording full time because he’s continuing elementary school in the U.S. now and has come back for the Christmas holidays. His seismometer station had a few problems at first: the wooden shelf he set it up on in a cupboard proved to be not very stable and it needed regular checking to make sure the arm was aligned with the magnet. Then some local quakes were missed because a plug had fallen out. Then when he copied the data files for viewing in Amaseis, he discovered somebody/something had pulled on the seismometer signal cable, knocking the pickup coil way out of alignment. Luckily though, the earthquake from Russia occurred several hours before that and the recording was very clean. But even then he couldn’t work out where it came from as the time didn’t match any of the earthquakes reported around the world. Fortunately he realized his computer was still set to Taiwan time, so we made allowance for that, then did some quick calculations, and the arrival time of the waves confirmed it was the one from Russia after all. His house is some distance from the road and I’m told the foundations go right down to bedrock, so it’s a good location for a seismometer. This is how he set up his station:
The cupboard doors, when closed are a convenient way to stop drafts from blowing the arm around.