The seismometer consists of a levitated graphite rod with a natural period of approximately 5 seconds. The rod is critically damped by means of eddy currents and its position is sensed by an LED and photo resistor. The device is therefore a displacement and tilt sensor. Data is sampled at 5 readings per second and logged with a time stamp for later analysis.
Two rows of Neodymium Iron Boron (NIB) cylindrical permanent magnets, magnetized across the diameter, are used to create a linear region of high field gradient. A half cylinder (1/8" diameter) of very pure graphite, which is strongly diamagnetic, will then levitate stably between them. By adjusting the length of the graphite rod the period can be adjusted. The graphite rod was delicately filed to leave approximately a half cylinder to give rotational stability so a flag, or shutter, could be attached to it and would stay pointing upwards. Critical damping was achieved by wrapping a small piece of aluminium cooking foil around the graphite. A further piece of aluminium foil was glued to the upper surface of the half cylinder to act as a moving light valve or shutter.
A small rectangular LED was positioned to one side of the graphite and a Cadmium Sulphide photoresistive cell on the other. The positions of these were adjusted so that the shutter foil half occluded the light from the LED from falling on the CdS cell.
The NIB magnets and graphite rod assembly, but the not the LED or CdS cell, were enclosed in a box made of microscope slides to protect them from air currents. See a diagram of all this. The whole setup is positioned on the floor of a basement garage in a Los Angeles apartment building with a cardboard box over it to protect it from light and air currents.
The LED is driven by a stabilised supply and the CdS cell has a resistance of around 300kOhms when the LED is half occluded by the shutter. The resistance of the cell is sampled at 5 cps by an HP/Agilent E1326B digital multimeter (DMM) at 5.5 digits precision and read into the computer via a GPIB (IEEE-488) bus. Reading the resistance directly is very wasteful of sensitivity and a Wheatstone bridge circuit could be used to remove the large constant component of the resistance. In this instance it doesn't matter since the 5.5 digit precision of the DMM ensures that the system's sensitivity is entirely limited by the microseism and local noise background. For simplicity, and to account for any missing data points, a time stamp in decimal seconds is logged for each data point and the data collection software writes a 9MB ASCII file of resistance and time readings every day. The system clock is corrected by a network time server every hour and timing accuracy is thought to be around +/- 0.25s.
An earlier, more economic, version of the data collection electronics used an HP 3478A DMM, which can be found on eBay for around $200. However one still needs a GPIB card for the computer and these also eBay at around $200. A much more economic solution would be to find an A/D converter card or analogue to serial external converter and read the data in through the serial port.
In the analysis software, data is passed through a 5'th order Butterworth bandpass filter whose high and low pass points are adjusted for differing distances of quakes and then through a 1 pole low pass filter that rolls over at 5s. This is to compensate for the loss of sensitivity to displacements slower than the resonant period of the graphite rod.