← Selected projects

Software development · Minebea Intec GmbH · June–July 2026

TConfig desktop application

A Windows application for monitoring and configuring industrial weighing instruments over RS232 and RS485.

My role

Application development & serial integration

Tools

Electron · React · TypeScript · Node.js

TConfig

Industrial instrument configuration

  1. Monitor
  2. Configure
  3. Calibrate
Windows · Electron · React
TConfig · application workflow overview

A clearer view of the instrument

I developed TConfig from an existing React interface into an Electron desktop application for real weighing instruments. Its monitoring, configuration, and calibration workflows connect an operator’s actions to the instrument’s actual state.

My work: Developed operator workflows, connected the interface to device communication, and added verified parameter-read behavior.

The engineering decisions

TCONFIG · V2.0.0

An interface can show an apparently successful change while the instrument holds a different value. Live weight frames, parameter replies, and calibration operations also share communication paths, so the application must distinguish them reliably.

  • Defined page workflows and checked device commands against protocol documentation and raw serial observations.
  • Separated editable drafts from confirmed instrument values, with serialized writes and readback comparisons where implemented.
  • Prioritized matching parameter responses ahead of live weight publication to keep replies from contaminating the weight display.
Design details & validation

Read without changing the instrument

Calibration-weight reads originally risked entering a control mode. Dedicated passive queries replaced that probing behavior, and concurrent reads were merged. Focused simulated-serial regression tests cover these read operations.

Make the serial boundary explicit

A preload/IPC bridge connects React to an Electron serial service. The service handles frame assembly, request queues, CRC validation, timeouts, and polling appropriate to the active page.

Recorded results

  • Delivered nine main application pages plus login and a Windows x64 installer with native serial support.
  • Added focused regression coverage for calibration reads; the handoff records seven passing tests and passing TypeScript checks.
  • Reduced the bundled font resource from 30.7 MB to approximately 209 KB through subsetting. This is a font-resource reduction, not an application-size claim.

Instrument feedback guided development. The final passive-read changes have focused simulated-serial coverage; the handoff does not include a subsequent complete hardware acceptance run.

I defined workflows, supplied and corrected protocol details, tested device behavior, and directed implementation iterations. AI coding tools supported implementation, refactoring, and test generation.

What the work delivered

Delivered nine main application pages and a Windows x64 installer. The handoff records seven passing tests for passive calibration-weight reads and passing TypeScript checks.

The project period refers to the internship. Release-package and handoff checks also include August–September 2026 records.