Spectrum Bench

A bench for the most-published data shape in chemistry and materials: a one-dimensional signal with peak structure. Drop a two-column ASCII export (.xy — whitespace-separated, headerless, # comment lines skipped) or a CSV/TSV whose first column is the abscissa and whose remaining columns are one trace each, then bind it with Add chart… as a Spectrum, which draws a connected trace rather than the cloud of points a scatter would give you. THE X AXIS IS THE THING TO GET RIGHT: NMR chemical shift and IR wavenumber are ALWAYS plotted high-to-low, so the technique buttons set the axis label and its direction together — pick IR and the axis runs 4000 down to 400 cm⁻¹ the way every published figure does; pick XRD, MS or UV-Vis and it runs the other way. Traces are sorted by x on import, so a file exported descending (which FTIR packages commonly do) still draws correctly and the direction stays the plate's choice rather than the instrument's. COMPARISON IS THE WORKFLOW, which is what the two overlay controls are for: Normalise scales every trace to its own tallest peak, without which a measurement in counts and a reference pattern in arbitrary units differ by orders of magnitude and the reference draws as a flat line along the axis; and a per-trace waterfall offset stacks a sample series so the traces read against each other instead of through each other. PEAKS: Pick takes the local maxima above a fraction of the trace's own full height — unit-free, so the same threshold works on counts, transmittance or relative abundance — keeps the strongest, and drops anything within one percent of the x span of a peak already kept. That separation rule is not a nicety: every sample on a strong peak's noisy flank is a local maximum, so without it a clean pattern reports as dozens of peaks stacked on one line. Labels that would overlap are dropped rather than overprinted. Zoom and pan act on the x window, which is how a spectrum is actually read, and the sidebar says where the window sits and how many peaks are marked. NOT READ: JCAMP-DX (.jdx/.dx), whose compressed ASCII encodings are a parser of their own, and no vendor binary — convert to two-column ASCII first. The bench does no baseline correction, no deconvolution and no curve fitting, so a marked peak is a local maximum and nothing more; integrated areas, fitted widths and assignments belong to the software that produced them.

Supported file formats: CSV, TSV, XY