University lab / Embedded data

Vibration,
sampled & analysed.

In a Dynamics lab, I used a microphone, amplifier and Raspberry Pi Pico to record a metal ruler's free vibration at 5 kHz. A Python workflow then validated a known tone, extracted vibration frequency, and estimated Young's modulus through a cantilever-beam model.

Raspberry Pi Pico Python Signal Processing Data Analysis
View workflow View results
A compact experimental chain: acquire the analogue response, turn it into a time series, then make an engineering judgement from it.
5 kHz sampling rate, giving a 2.5 kHz Nyquist frequency
499.77 Hz estimated from the known 500 Hz validation tone
238.81 GPa mean of the retained grouped Young's modulus estimates
Workflow

From an acoustic response to a material estimate

01 / ACQUIRE

Record a repeatable signal

Each 3 s recording captured the release and free decay of a ruler at 0.10 m, 0.15 m and 0.20 m overhangs.

02 / VALIDATE

Check the acquisition chain

A separate 500 Hz tone was processed with the same settings and returned 499.77 Hz before analysing the ruler trials.

03 / PROCESS

Make the time signal usable

Python removed the DC component, used a moving-average envelope to find the active window, and estimated frequency from zero crossings in the cropped response.

04 / INTERPRET

Connect frequency to stiffness

The fundamental frequency entered a cantilever-beam relationship with the measured geometry and mass per unit length to estimate Young's modulus.

Annotated physical experiment setup showing the ruler, microphone amplifier circuit and Raspberry Pi Pico
Report figure / acquisition setup The physical setup joined the cantilever ruler, microphone-amplifier circuit and Raspberry Pi Pico ADC in one measurement chain.
Python plot of the 500 Hz validation signal with detected zero crossings
Report figure / validation signal Using the same acquisition and Python workflow, the 500 Hz control signal was estimated at 499.77 Hz: a −0.046% error.
Results

Three lengths, nine trials, one evidence trail

Grouped measurements after the same signal-processing workflow was applied to each recording.
Overhang Retained trials Mean frequency Frequency SD Mean Young's modulus
0.10 m353.40 Hz0.11 Hz222.66 GPa
0.15 m324.35 Hz0.29 Hz234.38 GPa
0.20 m2 / 314.41 Hz0.29 Hz259.38 GPa
Chart of grouped Young's modulus estimates by ruler overhang length with error bars
Report figure / grouped result Grouped Young's modulus estimates from the retained trials; the error bars show the standard deviation within each overhang group.
What I learned

Embedded sampling is only the start

D1

A physical signal has to survive the hardware

I learned how microphone placement, amplifier gain, ADC sampling and the Nyquist limit affect whether a vibration measurement is useful at all.

D2

Python turns raw CSV into an argument

The useful work was not just calculating a frequency: it was centring the signal, identifying an energetic window, checking crossings, and applying the same method across trials.

D3

Data quality is part of the result

I documented clipped signals, 50 Hz interference and an 8.41 Hz outlier instead of treating every number as equally reliable. That made the final estimate more honest and more useful.

Reflection

Useful measurement, clear limits

The 500 Hz validation showed that the acquisition-and-analysis chain could recover a clean periodic signal. The ruler estimate, however, was sensitive to the real measurement conditions and simple beam assumptions.

Next iteration priorities

  • Reduce amplifier gain or increase microphone distance to prevent clipping before it reaches the frequency estimate.
  • Improve grounding and shorten wiring to reduce the visible 50 Hz electrical hum.
  • Use a stronger clamp and repeated micrometer measurements of thickness, which strongly influences the final modulus.
  • Compare the microphone result with an accelerometer or displacement sensor, and add more repeat trials to make outliers easier to identify.

Source note — This page condenses my CADE10002 Dynamics Lab Report. The values and limitations shown here are course-project evidence, presented as an engineering learning project rather than a production-grade material test.