Vibration, answered in firmware
On a station, the micro-vibration environment shows up directly in the surface quality of an FDM print. Classical active vibration control answers that with another actuator. This work does not: it modulates the print head's instantaneous kinematic speed at the firmware level, so the motion itself avoids exciting the structure's resonances. To validate it on a bench I built the disturbance — a single-axis shaker rig, mechanics through desktop software.
77TH INTERNATIONAL ASTRONAUTICAL CONGRESS · ANTALYA · 5—9 OCTOBER 2026 · INTERACTIVE PRESENTATION · ROLE: RESEARCH IDEA, CONTROL DESIGN, TEST RIG
WHAT THE WORK LOOKS AT
- How an ISS-like micro-vibration environment degrades FDM surface quality
- An FxLMS-based control approach against that disturbance
- A software answer instead of added hardware — no counter-actuator, no extra mass
- Speed modulation as the control handle: resonances are avoided, not cancelled after the fact
MY CONTRIBUTION
- The research idea, the experimental approach and the system concept
- Design of the control side
- Building the test rig the validation runs on
- Paper title: Real-Time Active Vibration Compensation in Microgravity Additive Manufacturing via Kinematic Speed Modulation
1 kHz
POSITION SET-POINT UPDATE
±150 mm
USABLE STROKE
~10 kg
TARGET LOAD CLASS
<200 USD
COST, EXCL. PRINTER & PC
WHAT I BUILT
- Mechanical layout and drive selection
- STEP/DIR driver chain and power electronics
- Firmware on an Arduino Nano
- Python/Tk control application and the test logic around it
WHY IT EXISTS
- The disturbance in the IAC study has to be physically reproducible to be argued about
- One project carries mechanics, electronics, firmware, desktop software and test procedure
- Open hardware — the whole rig is published, not described
- github.com/dereyurtali/single-axis-shaker-rig
SURFACE ROUGHNESS · CUSP FILTERING
Effects of process parameters on surface roughness in fused deposition modelling after cusp filtering
ALI DEREYURT · EBUBEKIR KOÇ
Layer height, wall count, infill density and print speed studied in a full factorial design: 324 surface profiles collected from 108 specimens, measured to ISO 4287/4288. To separate out the dominant effect of layer geometry I developed a cusp filtering method locked to the measured peak and valley positions. After filtering, print speed becomes significant at p = 0.005.
MULTI-OBJECTIVE OPTIMISATION
Multi-objective parameter optimisation in fused deposition modelling using machine learning and Bayesian optimisation
ALI DEREYURT · EBUBEKIR KOÇ
An end-to-end approach: the uploaded 3D model is read through four orthographic views and a vision-language model, roughness is predicted by monotonicity-constrained gradient boosting, and a Bayesian search balances roughness, material use, print time and structural strength — every candidate validated through the slicer. Shipped as a product at printimize.dereyurt.dev.
3DOIT ADDITIVE MANUFACTURING
ENGINEER CANDIDATE · AUG 2024 — FEB 2025
FDM production management for aerospace and industrial prototyping: printer-farm operation, parameter optimisation, DfM, dimensional verification and surface inspection.
FSMVÜ ALUTEAM
PART-TIME STUDENT · OCT 2021 — FEB 2022
EOS SLS and DMLS systems: build preparation, powder handling, depowdering and surface finishing; 3D scanning and part inspection on a Hexagon robotic arm.