Earthquake Science Fair Project

Building a Working Seismograph for Your Project

Most students grab an accelerometer module off Amazon, solder it to a breadboard, and call it done. The results look nothing like a real seismogram because they haven't accounted for how noisy cheap MEMS sensors are at low frequencies. I spent two weeks fixing this with a kid last year before we could get anything that resembled P-wave and S-wave arrival patterns on the serial plotter. The core problem is that acceleration-based sensors drift. You apply a constant tilt and the reading slowly climbs or falls on its own. That drift completely masks the tiny vibrations you're trying to capture. The workaround is either passive filtering through the mechanical setup or software-level high-pass filtering after calibration.

Earthquake Science Fair Project

A proper project uses a simple mass-spring system as the mechanical front end. A steel washer or small nut slides along a bent paperclip or hangs from a rubber band inside a shoebox. That moving mass carries a stylus—a pencil tip or a paintbrush bristle—that drags across a strip of paper mounted on a rotating drum. The drum is driven by a slow stepper motor or, if you want to keep the cost down, a manual crank. When you shake the base, the mass tries to stay stationary due to inertia, and the paper moves under it. The resulting trace is your seismogram. The trick nobody tells you is the length of the paperclip arm. Too stiff and the natural frequency is too high—you miss the low-frequency shaking. Too loose and the mass bounces around randomly. A #2 paperclip bent into a horizontal cantilever about three inches long with a one-inch overload gives roughly a 1-2 Hz resonant frequency, which is close to what you need for tabletop simulation. I learned this the hard way. My first build used a stiff spring and the trace looked like white noise. Switched to the paperclip and got clean sinusoidal traces when tapping the box. The difference was entirely the mechanical resonance matching the excitation frequency.

Software Approach If You Want Digital Data

If your project needs a digital readout, hook a triple-axis accelerometer like the ADXL345 to an Arduino. Set it to 10-bit mode and sample at 100 Hz. Export the data to a .csv file and plot it in any spreadsheet. What most people miss is that you need to subtract the mean and apply a high-pass filter at 0.5 Hz to remove the gravity component that shows up when the board tilts even slightly. The high-pass filter is the step that turns garbage into something believable. Without it, your baseline drifts across the entire plot area and any real signal gets lost. A simple first-order high-pass in Processing or even Excel works fine for a science fair. A second-order Butterworth is better but usually overkill at this level.

Simulating Earthquake Waves

You don't need an actual earthquake. A rubber mallet hitting the side of the box simulates P-waves. Shaking the base laterally simulates S-waves. Surface waves are harder to fake on a desk but you can approximate them by dragging the box back and forth while it's recording. The time gap between the mallet strike trace onset and the lateral shake onset is your P-S time difference, which is exactly what teachers look for in the analysis section. Record at least three trials for each input type. The variance in arrival times matters more than any single reading. My best students always had the tightest clustering across repeated hits. That's where the real points are.

Calibration Is Non-Negotiable

Before the fair demo, calibrate the system with a known input. Drop a small weight from a measured height onto the box and note the peak acceleration. That gives you a reference point. Without it, your amplitude claims are just guesses. Judges will ask for the calibration method and they will mark you down if you don't have one. The piezoelectric version I mentioned earlier is fine for demonstrating the concept, but the mass-spring stick-trace design produces the kind of output that looks exactly like a real seismogram. That visual match is what wins ribbon placements. It's also cheaper and requires zero programming.

Common Pitfalls

Don't use a smartphone app as your primary sensor unless the rubric specifically allows it. The sampling rate is usually too low and the built-in filtering removes the very data you're trying to show. Don't tape the paper down too tightly—friction damps the trace. Don't skip the enclosure; ambient vibration from the classroom floor ruins everything below 5 Hz. If your mass-spring system is overshooting and oscillating for more than five seconds after a single tap, your damping is too low. A small piece of foam between the mass and the frame or a drop of silicone grease on the paperclip contact point reduces oscillation enough to get clean successive traces without killing the amplitude.

What to Show on Display Board

Include the seismogram trace alongside a labeled diagram showing the fault line, epicenter distance, and the P-S interval. Calculate the distance using the standard 8 km/s S-minus-P rule. Show your calibration data in a small table. That's it. No need for elaborate animations or animated GIFs. The actual paper trace with real data beats a perfectly rendered simulation every time. One last thing. Test everything twice in the week leading up to the fair. Paper slips. Styluses wear down. Springs lose tension. I've seen projects fail at the judging table because the mass was stuck to the frame from humidity. Keep spares and know how to adjust the paperclip length on the fly.