Lehman seismometer

The 1980s and 90s were an active period for earthquakes in northern California, where I grew up. I have vivid memories of the 1989 Loma Prieta earthquake, among others. At the time, I was in high school and read on USENET about a home-made seismograph described in a 1979 Amateur Scientist column in Scientific American magazine. The article explained how to construct an instrument from pipe fittings, piano wire, and other simple parts, capable of recording large earthquakes around the world. I immediately knew I had to build one!

(I was lucky to have USENET access at a time when its readers were mostly academics and tech workers. You could ask a question and likely get an answer from someone at a national lab or major university. I ended up getting advice for the seismograph from an employee at JPL! Similarly, I exchanged dozens of letters about magnetometer circuits with an engineer at Brookhaven National Lab. This was old-fashioned, time-consuming paper correspondence, as it was too difficult to send drawings over email. It still amazes me that people would spend that kind of time mentoring an anonymous high-school kid, and I wonder if it's even possible in today's work environment. It might help our so-called STEM crisis.)

The essence of the Lehman seismometer is a nearly horizontal, critically damped pendulum with a natural free period of about 15 seconds. The structure resembles a swinging gate held by a support wire, thus it is sometimes called a “garden gate seismometer.” The damping can be provided by a magnet (as in the article) or an old credit card partially submerged in oil. This broadens the frequency response somewhat; without it, the undamped pendulum would simply ring at 1/15 Hz. In the article, movement is measured with a coil moving in a magnet gap; today one might try an optical method instead with a laser pointer reflecting from a mirror glued to the boom, and some kind of differential light sensor. The whole thing is then enclosed in some kind of box to prevent disruption from air currents; I used some old printer enclosures, relics from the days of noisy daisy-wheel and dot-matrix technology.

I remember being most proud of the home-brew 8-bit digitizer I used to record the signal. It consisted of an Analog Devices A/D and a Motorola 68230 “Parallel Interface and Timer” hand-soldered onto an ISA-bus prototyping card. This taught me a lot about microprocessor busses, address decoding, and hardware interrupts! The code was all custom C under MS-DOS, and at one point even included the ability to auto-detect and save events using deviations from the running averages of different Fourier-transform bins.

This type of design can be immensely educational, despite not having a very flat or wideband frequency response. Most people are surprised how sensitive such a simple instrument can be. Mine was capable of detecting magnitude-6.5 or larger quakes over most of one hemisphere, and magnitude-5.5 or larger anywhere in the US, Canada, and Mexico. It even registered one of the last underground nuclear tests in Nevada! It is often possible to distinguish P- and S-wave arrivals and thereby estimate the great-circle distance to the epicenter.

Modern seismographs use negative force feedback, just like good electronic amplifiers, to cover a much larger bandwidth with good linearity. Amateurs are building nice examples of these instruments too. I think one of them may be in my future...

school closet site top view