Snapshots from the powerade_blue_1120 continuous-flow run on the LIGHTNING-27 antenna:
the test setup, the raw capture, and the trained model. Each opens as a self-contained page —
all plots and images are embedded, nothing is fetched from elsewhere.
Test setup
Test setup: continuous-flow dilution rig
How the measurement is run end to end -- the continuous-flow dilution rig that takes a sample from its starting concentration down to zero while the antenna sweeps, and how the scale weight is turned into the concentration label at each moment. Read this first: it is the physical context the two snapshots below are measuring.
Open →
Capture
orbio-py capture snapshot
A frozen view of the orbio-py capture GUI for the powerade_blue_1120 run (LIGHTNING-27 antenna): the live antenna spectra and the History panel, with the entered (scale-derived) concentration and the model's per-sweep estimate over the whole run. This is the raw signal side -- what the device actually saw as the sample diluted.
Open →
Model
orbio-ml-py training snapshot
A frozen view of the orbio-ml-py training GUI for the same powerade_blue_1120 run: the lab data summary, the trained-model results (RMSE / R2 per method and pipeline), and the true-vs-predicted plots. This is the model side -- how well concentration is recovered from the spectra in the capture snapshot.
Open →