Aayushya Patel

Feature · 2024 —

A six-axis arm for cold plasma wound treatment

A hand-held plasma probe cannot hold a consistent standoff over an irregular wound surface, so the delivered dose is uneven and outcomes vary.

That is a motion-control problem wearing a medical coat.

San José State University, Dept. of Mechanical & Aerospace Engineering · under Dr. Sohail H. Zaidi

Degrees of freedom
6
Accuracy deviation
< 0.1 mm over extended paths
Positional error
0.027 mm mean single axis, 50 positions across 0–50 mm, caliper
Joint torque
~10 Nm at the shoulder (65 Ncm motor, 20:1 planetary); reduction varies per joint
End-effector payload
0.5 kg tested, 1.0 kg by FEA
Thermal cutoff
40 °C, enforced by the host safety arbiter
Structure
PETG, printed in-house

01 The system working

The arm holding the DBD torch at controlled standoff while the discharge runs onto a tissue phantom. Everything after this chapter is an answer to how it holds that distance.

Fig. 1 Twelve seconds out of a twenty-six second treatment run. The torch is under path control, not being held still — the standoff is what stays constant, not the position.
Fig. 2 Full-resolution frame from the same run. The glass tube is the discharge electrode; the gap between its tip and the phantom is the entire problem this machine exists to solve.
Fig. 3 Six frames sampled across the run. Read left to right, top to bottom: the torch head displaces across the phantom rather than dwelling, which is what turns a point dose into a treated area.
Fig. 4 The bench mid-run. CAD assembly open on the laptop, oscilloscope monitoring the discharge, phantom under the torch, lab posters behind.

02 Electronics and mechanics

One board, six axes, and a set of printed links. All of it designed, laid out, and assembled here rather than bought as a kit, which is the only reason the failure modes were mine to fix.

Fig. 5 The control board: six TMC5160 Pro stepper drivers under the red heatsinks, a Teensy 4.1, bulk power regulation, magnetic encoder inputs, and the hardware emergency stop. One EasyEDA project, routed from scratch.
Fig. 6 The assembled arm alongside that board, wired and powered. Printed PETG links. Reduction varies per joint: ~10 Nm at the shoulder through a 20:1 planetary, 4:1 belts at the wrist. 1 kg near the base, 0.5 kg across the full workspace.
Fig. 7 Every joint module, stepper, and the driver board laid out. Every structural part in this photograph was printed rather than purchased. The control board, the sensor head, and the path planner are my own design.
Fig. 8 A single joint: printed housing, integrated ring gear, encoder bore, and ventilation for a stepper that has to survive continuous duty.
Fig. 9 The mechanical design behind Fig. 8 — belt-driven joint, bearing stacks, ribbed base, fastener stack-up.

03 Plasma and diagnostics

The arm is only half of it. The other half is knowing what the discharge is actually doing, which means the discharge has to be measured and not assumed.

Fig. 10 Dielectric barrier discharge in the chamber. The violet emission is excited nitrogen and oxygen species — the same species that do the antimicrobial work, which is why the light is a diagnostic and not just a photograph.
Fig. 11 The torch firing in a darkened lab with an Ocean Optics HR4C4660 acquiring on the laptop, process gas cylinder behind. The spectrum is how the species in Fig. 10 stop being a guess.
Fig. 12 The full diagnostic bench: CTP-2000K generator with live voltage and current, oscilloscope for discharge characterisation, spectrometer laptop, gas supply.
Fig. 13 The chamber on its own — glass tube, internal ring electrodes, gas feed, HV leads.

04 What it looked like before

Kept in on purpose. A portfolio that only shows the finished machine is a claim about taste, not about work.

Fig. 14 The arm moving before the control loop was tuned: twitchy, overshooting, unusable. Everything in chapter 02 exists because of what this clip shows.
Fig. 15 And where it started — red printed arm, stepper drivers loose on a breadboard, jumper wires. Every part of this was replaced.

Reviewed

Archive The complete record — walkthrough, every plate, both raw clips, the papers Back to the front page