Atomionics is a mid-sized Singaporean startup using quantum sensors for gravitational data to explore underground. However, their sensor had excessive noise, partially due to vibrations that affected the gravity measurements. I was tasked with determining whether a noise-cancellation system could improve the data and implementing a solution.

This vibration isolation platform reduced force attenuation from 1×10^-5 to 3×10^-7 m/s², decreasing noise from ground and acoustic vibrations. In drive tests, the platform continued functioning after an impact over 90% of the time. Even after substantial impacts, fine-tuning restored performance within 1 minute. The platform exhibited minimal flexing and tilting, verified through laser alignment and the tunable sliding system. Using this damping platform, I confirmed the company could collect gravity data with uncertainty reduced by an order of magnitude.

This project requires analyzing the quantum system and how the external noise will influence the measuring results. I simulated the system with external disturbance under different profiles and quantified the result uncertainty levels to do this. From this, I deduce the system required to filter out the noise and approach the problem from a product design perspective to design a robust system. These are the steps and the design decisions I made:

1. Varifying noise source and contribution
   I collected background noise, performed simulations on different noises, and incorporated the quantum detection scheme. It is shown that more than 80% of the noise of our data is from the ground vibration noise.Simulation of different close-loop attenuation profile is done to decide how good the attenuation need to be.
2. Design requirements and decisions
   Besides the attenuation profile, other design requirements are considered. The device has to be portable, impact-withstanding, and has a low deflection angle for the mirror mounted on it. A voice coil actuation schematic is selected to implement a low-lag feedback loop. The voice coil is fine-tunable by optics stages. Slots are implemented to allow individual movement of the supporting springs to ensure flatness.
3. Control scheme
   The noise bottlenecks are strict in order to achieve the accuracy I want. A mix of high-frequency DAQ hardware and precise current amplifiers is selected. The control system incorporates frequency domain and lead-lag controller design. All the hardware and plants are modeled on Matlab and implemented through Lua scripts.